wfp/ifpri impact evaluation of cash, food vouchers, … · cre cruz roja ecuatoriana (ecuadorian...
TRANSCRIPT
WFP/IFPRI
Impact Evaluation of Cash, Food Vouchers, and Food Transfers among
Colombian Refugees and Poor Ecuadorians in Carchi and Sucumbíos
Final Report
Melissa Hidrobo
John Hoddinott
Amy Margolies
Vanessa Moreira
Amber Peterman
International Food Policy Research Institute
Poverty, Health, and Nutrition Division
2033 K Street, NW
Washington, DC 20006
USA
FIRST DRAFT: May 4, 2012
CURRENT DRAFT: October 17, 2012
iii
Table of Contents
Acronyms ................................................................................................................................................... ix
Executive Summary .................................................................................................................................. xi
1. Introduction ............................................................................................................................................ 1
1.1 Background ....................................................................................................................................... 1
1.2 Ecuador Context and Study Objectives ........................................................................................ 2
2. Intervention ............................................................................................................................................. 4
2.1 Site Selection and Beneficiaries ...................................................................................................... 4
2.2 Intervention ....................................................................................................................................... 5
3. Evaluation Design .................................................................................................................................. 8
3.1 Study Design ..................................................................................................................................... 8
3.2 Study Sample .................................................................................................................................... 9
3.3 Estimation Strategy .......................................................................................................................... 9
4. Data ........................................................................................................................................................ 12
4.1 Survey Instruments and Topics ................................................................................................... 12
4.2 Attrition Analysis ........................................................................................................................... 13
4.3 Baseline Analysis (by Treated and Comparison) ...................................................................... 15
5. Experience with Program ................................................................................................................... 20
5.1 Experience with Program ............................................................................................................. 20
5.2 Nutrition Trainings ........................................................................................................................ 28
5.3 Nutrition Knowledge .................................................................................................................... 28
6. Impact on Food Consumption and Dietary Diversity .................................................................... 31
6.1 Indicators and Descriptive Statistics ........................................................................................... 31
6.2 Impacts of Pooled Treatment on Food Consumption and Dietary Diversity ....................... 34
6.3 Impact by Treatment Arm on Food Consumption and Dietary Diversity ............................ 37
6.4 Impacts by Food Groups ............................................................................................................... 39
6.5 Impacts by Nationality .................................................................................................................. 43
6.6 Impacts on Nonfood Expenditure ............................................................................................... 45
7. Impact on Social Capital: Trust, Discrimination, and Participation ............................................. 48
7.1 Baseline and Follow-Up Descriptive Statistics .......................................................................... 48
7.2 Impacts on Trust, Discrimination, and Community Participation ......................................... 50
7.3 Impacts on Individual Questions................................................................................................. 52
iv
7.4 Impacts by Nationality .................................................................................................................. 55
8. Anemia .................................................................................................................................................. 57
8.1 Indicators and Descriptive Statistics ........................................................................................... 57
8.2 Impact of Treatment on Anemia Indicators for Children 6-59 Months Old .......................... 59
8.3 Impact of Treatment on Anemia Indicators for Adolescent Females ..................................... 62
9. Gender Issues........................................................................................................................................ 65
9.1 Decisionmaking .............................................................................................................................. 65
9.2 Intimate Partner Violence ............................................................................................................. 73
10. Costing and Cost-Effectiveness ........................................................................................................ 78
10.1 Costing Methods and Results ..................................................................................................... 78
10.2 Cost-Effectiveness ........................................................................................................................ 81
11. Qualitative Lessons on Nutrition Training, Knowledge, and Behavior Change ...................... 83
11.1 Background ................................................................................................................................... 83
11.2 Methodology ................................................................................................................................. 84
11.3 Results ............................................................................................................................................ 85
11.4 Qualitative Conclusion ................................................................................................................ 94
12. Impact Evaluation Conclusion ......................................................................................................... 96
Appendix 1: WFP Poster for Supermarkets ......................................................................................... 99
Appendix 2: Costing Methods and Definition ................................................................................... 100
Appendix 3: Detailed Cost Tables ....................................................................................................... 104
References ............................................................................................................................................... 118
v
List of Tables
4.1 Modules in the baseline household questionnaire .................................................................... 13
4.2 Attrition rates .................................................................................................................................. 14
4.3 Attrition rates, by treatment and comparison groups .............................................................. 14
4.4 Difference in attrition rates, by treatment arms ......................................................................... 14
4.5 Differential attrition analysis (mean values of baseline characteristics) ................................ 16
4.6 Baseline characteristics, by treatment and comparison arms .................................................. 17
4.7 Baseline characteristic, by treatment arm ................................................................................... 19
5.1 Time and cost to receive the transfer, by province .................................................................... 20
5.2 Time and cost to receive transfers, by modality ........................................................................ 21
5.3 Transparency and administrative processes, by province ....................................................... 22
5.4 Transparency and administrative processes, by modality ....................................................... 22
5.5 Satisfaction with transfer modality, by treatment status .......................................................... 23
5.6 Difficulties experienced by voucher beneficiaries, by province .............................................. 24
5.7 Difficulties experienced by food beneficiaries, by province .................................................... 24
5.8 Difficulties experienced by cash beneficiaries, by province ..................................................... 25
5.9 Decision on how to spend or use the transfer in male headed households, by
modality ........................................................................................................................................... 26
5.10 Decision on how to spend or use the transfer in female-headed households, by
modality ........................................................................................................................................... 26
5.11 Reported uses of last food transfer, by province ....................................................................... 27
5.12 Reported uses of last cash transfer, by province ........................................................................ 27
5.13 Reported use of vouchers, by province ....................................................................................... 28
5.14 Nutrition training sessions ............................................................................................................ 28
5.15 Baseline means, by pooled treatment .......................................................................................... 29
5.16 Follow-up means, by pooled treatment ...................................................................................... 30
6.1 Aggregate food groups and weights to calculate the Food Consumption Score .................. 32
6.2 Baseline food consumption and dietary diversity measures, by treatment status ............... 33
6.3 Follow-up food consumption and dietary diversity measures, by treatment status ........... 33
6.4 Impact of pooled treatment on food consumption measures, with and without
covariates ......................................................................................................................................... 35
6.5 Impact of pooled treatment on dietary diversity measures, with and without
covariates ......................................................................................................................................... 36
vi
6.6 Impact of treatment modalities on food consumption measures ............................................ 37
6.7 Impact of treatment modalities on dietary diversity measures ............................................... 38
6.8 Impact of treatment on poor to borderline food consumption ................................................ 39
6.9 Impact of pooled treatment on food frequency, by food groups ............................................ 41
6.10 Impact of treatment arms on food frequency, by food groups ................................................ 41
6.11 Impact of treatment arms on log per capita caloric intake, by food groups .......................... 42
6.12 Impact of treatment arms on the monetary value of log per capita food consumption,
by food groups .............................................................................. Error! Bookmark not defined.
6.13 Differential impact on food consumption with respect to nationality, by pooled
treatment.......................................................................................................................................... 43
6.14 Differential impact on food consumption with respect to nationality, by treatment arms . 43
6.15 Differential impact on dietary diversity with respect to nationality, by pooled
treatment.......................................................................................................................................... 44
6.16 Differential impact on dietary diversity with respect to nationality, by treatment arms .... 44
6.17 Impact of pooled treatment on household expenditure (logs) ................................................ 46
6.18 Impact of treatment arms on household expenditure (logs) .................................................... 46
6.19 Impact of treatment arms, by nationality, on household expenditure (logs) ........................ 47
7.1 Baseline means, by pooled treatment .......................................................................................... 49
7.2 Follow-up means, by pooled treatment ...................................................................................... 49
7.3 Impact of pooled treatment on social capital measures ............................................................ 51
7.4 Impact of treatment modalities on social capital measures ..................................................... 52
7.5 Impact of treatment modalities on trust indicators ................................................................... 53
7.6 Impact of treatment modalities on discrimination indicators ................................................. 54
7.7 Impact of treatment modalities on community participation .................................................. 55
7.8 Differential impact with respect to nationality, pooled treatment .......................................... 56
7.9 Differential impact with respect to nationality, by treatment arms ........................................ 56
8.1 Anemia classifications ................................................................................................................... 57
8.2 Baseline anemia measures among 6–59 month old children, by pooled treatment ............. 58
8.3 Follow-up anemia measures among 6–59 month old children, by pooled treatment .......... 58
8.4 Baseline anemia measures among adolescents aged 10 – 16, by pooled treatment .............. 58
8.5 Follow-up anemia measures among adolescents aged 10 – 16, by pooled treatment .......... 59
8.6 Impact of pooled treatment on anemia measures among 6–59 month old children ............ 59
8.7 Impact of pooled treatment on anemia measures among 6–59 month old children
with covariates ................................................................................................................................ 60
vii
8.8 Impact of treatment modalities on anemia measures among 6–59 month old children
with covariates ................................................................................................................................ 61
8.9 Differential impact on anemia measures among 6 - 59 month old children with
covariates with respect to ethnicity, by treatment arms ........................................................... 61
8.10 Impact of pooled treatment on anemia measures among adolescent girls aged 10–16 ....... 62
8.11 Impact of pooled treatment on anemia measures among adolescent girls aged 10–16
with covariates ................................................................................................................................ 63
8.12 Impact of treatment modalities on anemia measures among adolescent girls
aged 10-16 covariates ..................................................................................................................... 64
8.13 Differential impact on anemia measures among adolescent girls aged 10–16 with
covariates with respect to ethnicity, by treatment arms ........................................................... 64
9.1 Baseline means of decisionmaking indicators, by pooled treatment ...................................... 67
9.2 Follow-up means of decisionmaking indicators, by pooled treatment .................................. 68
9.3 Impact of pooled treatment on sole or joint decisionmaking .................................................. 69
9.4 Impact of pooled treatment on sole or joint decisionmaking, with covariates ..................... 70
9.5 Impact of pooled treatment on disagreements regarding decisionmaking ........................... 71
9.6 Impact of pooled treatment on disagreements regarding decisionmaking, with
covariates ......................................................................................................................................... 71
9.7 Impact of treatment modalities on sole or joint decisionmaking, with covariates ............... 72
9.8 Impact of treatment modalities on disagreements regarding decisionmaking
domains with covariates ............................................................................................................... 72
9.9 Intimate partner violence baseline means, by pooled treatment ............................................. 74
9.10 Intimate partner violence follow-up means, by pooled treatment ......................................... 74
9.11 Impact of pooled treatment on intimate partner violence measures ...................................... 76
9.12 Impact of treatment modalities on domestic violence measures with covariates................. 77
9.13 Impact of treatment by gender on domestic violence measures with covariates ................. 77
10.1 Modality-specific cost of improving food security outcomes by 15 percent ......................... 82
A3.1 Detailed budgeted costs, by modality ....................................................................................... 104
A3.2 Detailed actual costs, by modality ............................................................................................. 108
A3.3 Modality specific costs (based on actual costs) ........................................................................ 113
A3.4 Non-modality specific costs (based on actual costs) ............................................................... 115
viii
List of Figures
2.1 Map of intervention provinces ....................................................................................................... 4
3.1 Illustration of the double-difference estimate of average program effect .... Error! Bookmark
not defined.
4.1 Composition (by shares) of monthly household expenditure ................................................. 18
6.1 Density graphs of food consumption and dietary diversity indices, by treatment and
control arms at baseline and follow-up ....................................................................................... 34
6.2 Percent increase in food security, by treatment arms ............................................................... 38
7.1 Percent change in social capital indicators, by treatment arms ............................................... 52
10.1 Total cost to transfer $40, by modality ........................................................................................ 79
10.2 Modality-specific and nonspecific costs ...................................................................................... 80
10.3 Total cost by type (percent), by modality ................................................................................... 81
ix
Acronyms
ABC-I Activity-based Costing Ingredients
ATM Automatic Teller Machine
BDH Bono de Desarrollo Humano (Human Development Grant)
CCT Conditional Cash Transfer
CEPAR Centro de Estudios de Población y Desarrollo Social
CO Country Office
CRE Cruz Roja Ecuatoriana (Ecuadorian Red Cross)
CSPro Census and Survey Processing System
DDI Dietary Diversity Index
DHS Demographic and Health Survey
EFSA Emergency Food Security Assessment
FCS Food Consumption Score
FGD focus group discussions
GoE Government of Ecuador
HDDS Household Dietary Diversity Score
HIAS Hebrew Immigrant Aid Society
IFPRI International Food Policy Research Institute
NGOs nongovernmental organizations
PCA principal components analysis
PMA Programa Mundial de Alimentos (World Food Programme)
PRRO Protracted Relief and Recovery Operation
UNHCR United Nations High Council for Refugees
WFP World Food Programme
xi
Executive Summary
1. This report is the final impact evaluation of the World Food Programme’s Food, Cash, and
Voucher intervention and contains analysis on outcomes including food security, social
capital, anemia, and gender issues. Due to the targeting of Colombian refuges and poor
Ecuadorians in Northern Ecuador, it also examines whether program impacts varied by
nationality. In addition, to the impact assessment, this report provides evidence
surrounding participants experience with the program, conducts a costing study to examine
which modality (food, cash, or voucher) provides the greatest benefit for the amount of
funds invested, and presents the results from a qualitative study on the efficacy of the
nutrition trainings that accompanied the transfers.
2. Chapter 2 describes the Food, Cash, and Voucher intervention. It explains how beneficiaries
were selected and the different components of the program.
3. Chapter 3 presents the methods used in this study. It explains the rationale behind our use
of randomization and ANCOVA impact estimates.
4. Chapter 4 describes the data used in this study. It provides information on the survey
instruments and the study sample at baseline and follow-up. It conducts an attrition
analysis and a baseline analysis of characteristics across treatment and comparison
households and finds that randomization was relatively successful at balancing baseline
characteristics.
5. Chapter 5 describes participant’s experience with the program and the nutrition knowledge
gained from the nutrition sessions. The main findings are as follows:
The cash transfer incurs the lowest costs to participants in terms of waiting times and
transportation costs.
A higher percentage of participants prefer to receive all the transfer in cash as oppose to
all in food and voucher.
The transparency of the program and administrative process are very clear to
participants.
The main complaints of voucher recipients are lack of food items and higher prices in
supermarkets. The main complaint of food recipients is torn food packaging, and the
main complaint of cash recipients is lack of understanding on how to use the debit
cards.
Across all three modalities the transfers are reported to be mainly used for consumption
of food items; however, voucher recipients in comparison to cash recipients spend a
larger percentage on food. Almost none of the food transfer or voucher is sold to buy
other items. Besides food consumption, food recipients tend to share their transfer with
friends or family, or save their transfer for later use. Cash recipients also report saving a
small share of their cash for later use and spending a small portion on nonfood items.
xii
Nutrition knowledge increases from baseline to follow-up with the largest knowledge
gains occurring on food items that are rich sources of iron and vitamin A, and on
feeding practices for infants.
6. Chapter 6 presents the impact results of the Food, Cash, and Voucher program on food
security. The main findings are the following:
Impact of combined transfer program on food security outcomes: Overall, program
participation leads to large and significant increases across a range of food security
measures, with the value of per capita food consumption increasing by 13 percent, per
capita caloric intake increasing by 10 percent, Household Dietary Diversity Score
(HDDS) improving by 5.1 percent, Dietary Diversity Index (DDI) by 14.4 percent, and
Food Consumption Score (FCS) by 12.6 percent.
Impact by treatment arms on food security outcomes: All three modalities significantly
increase the value of food consumption, caloric intake, and dietary diversity as
measured by the DDI, HDDS, and the FCS. However, the increase on dietary diversity
measures is significantly larger for the voucher modality and the increase in caloric
intake is significantly larger for the food modality.
Impact of treatment arms on food group consumption: The food modality leads to
significant increases in the number of days a household consumes 5 out of 12 food
groups: cereals; roots and tubers; meat and chicken, fish and seafood; and pulses,
legumes, and nuts. The cash transfer leads to significant increases in 7 food groups: roots
and tubers; vegetables; meat and poultry; eggs; fish and seafood; pulses, legumes, and
nuts; and milk and dairy. Finally, the voucher leads to significant increases in 9 food
groups: cereals; roots and tubers; vegetables; fruits; meat and poultry; eggs; fish and
seafood; pulses, legumes, and nuts; and milk and dairy. The impact of vouchers on the
frequency of consumption is significantly different to that of food transfers for
vegetables, eggs, and milk and dairy.
Impacts by nationality: Both Colombians and Ecuadorians benefit from participating in
the program; however, Colombians in the food and cash group experience significantly
greater gains in dietary diversity measures than Ecuadorians.
Impacts on nonfood expenditures: With the exception of small increases in expenditure
on toys, the combined transfer program does not lead to increases in nonfood
expenditures.
7. Chapter 7 presents the impact results of the program on social capital (trust, discrimination,
and group participation). The main findings are the following:
Impact of combined transfer program: Program participation significantly increases
participation in groups and it significantly decreases the probability of experiencing
discrimination.
xiii
Impact by treatment arms: Although only cash leads to a significant decrease in
discrimination, the size of the impact is not different across treatment arms. Cash also
leads to a significant increase in trust of institutions and decrease in trust of individuals.
The increase in trust of institutions is significantly different to that of vouchers and the
decrease in trust of individuals is significantly different to that of food. On the other
hand, only vouchers lead to significant increases in participation in groups, and the size
of the impact is significantly different to that of cash.
Impacts by nationality: Increases in participation in groups are significantly higher for
Colombians than Ecuadorians and this is mainly due to significant differential impacts
from the cash arm.
8. Chapter 8 describes the data on anemia and the indicators that are used to conduct the
impact analysis for children aged 6 to 59 months and for adolescent girls aged 10 to 16. The
main findings from the impact analysis are the following:
Impact of combined transfer program: Overall participation in the program does not
lead to significant changes in hemoglobin levels or anemia classifications for either
young children or adolescent girls.
Impact by treatment arms: The program does, however, lead to a significant decrease in
hemoglobin levels and increase in anemia for children in the food group, and this
impact is significantly different from that of the voucher group. The program also leads
to a significant increase in moderate and severe anemia for children in the cash group.
For adolescent girls, there are no significant impacts across modalities.
Impacts by nationality: When interacted with nationality, the impact of food on
hemoglobin and anemia of children is significantly larger among Colombian
households.
9. Chapter 9 describes the data on women’s empowerment and the indicators for household
decisionmaking and intimate partner violence that are used to conduct the impact analysis.
The main findings from the impact analysis are the following:
Impact of combined transfer program: Overall, transfers lead to a significant decrease in
intimate partner violence; however, there is no impact on decision-making indicators.
Impact by treatment arms: The food treatment arm leads to a significant impact on
experience of disagreements regarding child health. Otherwise, there are no significant
impacts by treatment arm for women’s decisionmaking. While all three treatment arms
lead to significant decreases in physical/sexual violence, only cash and food lead to
significant decreases in controlling behaviors. However, there are no significant
differences across modalities in the size of the impact for any of the intimate partner
violence indicators.
xiv
Impacts by nationality: There are no significant differential impacts with respect to being
Colombian on any of the household decisionmaking indicators or intimate partner
violence indicators.
10. Chapter 10 presents the results from the costing study and finds that the food modality is
the most costly, and the cash and voucher less costly. In terms of cost-effectiveness across
food security outcomes, vouchers are the least costly means to improve these outcomes
while food is the most costly means to improve these outcomes.
11. Chapter 11 presents the findings from the qualitative work which generally support and
triangulate the results of the quantitative study. Beneficiary opinions of the nutrition
trainings are positive, and there seems to be some evidence of behavior change, particularly
in the testing and use of recipes as well as food purchase patterns. However, it is not clear
whether this effect will persist without the added benefit of the transfer. In addition, the
inclusion or involvement of spouses in the nutrition training and behavior change process
may lead to more favorable overall outcomes.
1
1. Introduction
1.1 Background
Developing country governments and donors are increasingly interested in moving
away from commodity-based assistance, such as food aid, and replacing it with alternative
transfer modalities such as cash and vouchers. In theory, cash is preferable to in-kind transfers
because it is economically more efficient (Tabor 2002). In addition, cash does not distort
individual consumption or production choices at the margin (Subbarao, Bonnerjee, and
Braithwaite 1997). Provided certain assumptions hold, cash transfers provide recipients with
freedom of choice to make the most needed expenditures, including human capital investments,
and give them a higher level of satisfaction at any given level of income than is the case with
food (Hanlon, Barrientos, and Hulme 2010). Cash distribution can also stimulate agricultural
production and nonagricultural activities by shifting out the demand curve for these items.
Further, distributing cash is likely to be cheaper than distributing food or other commodities.
For example, in-kind administrative costs can be 20-25 percent higher than that of cash transfers
(Cunha 2010; Ahmed et al. 2009, 2010).
The literature on the use of alternatives to food transfers has been summarized in papers
by Gentilini (2007) and Sabates-Wheeler and Devereux (2010). Both note that in contrast to the
heated debates regarding the use of alternatives to food, a more careful examination of the
issues suggests that both have benefits and drawbacks. In terms of their impact on beneficiaries,
the impact of a food transfer compared to one received through an alternative modality depend
upon at least six factors:
In the case of a food transfer, is the size of the transfer “inframarginal” (less than what
the household would have consumed without the transfer) or “extramarginal” (greater
than the amount of that commodity the household would have consumed without the
transfer)?
In the case of a food transfer, is the food product a “normal” (quantity consumed rises
when household income rises) or “inferior” good (quantity consumed falls when
household income rises)?
The net value of the transfer to the household after all transactions costs are taken into
account. Examples that affect this value include
o If the household sells some of the food transfer, the price they receive for that
transfer relative to the value of the transfer at current market prices;
o If the household sells a voucher, whether they receive the full value of the
voucher or whether it is sold at a discount;
o The costs of going to markets and using vouchers and/or cash to purchase food
and other goods.
The extent to which a food transfer or alternative modality is associated with the
perceived obligation to use this transfer in a particular fashion (for example, food
2
vouchers “should” be used to purchase food; food transfers “should” be shared with
extended family members).
The interaction between the transfer modality and the gender of the recipient. For
example, if food and food transfers are a “woman’s” resource, while cash and cash
transfers are a “man’s” resource, then differences in preferences between men and
women may result in different uses of transfers obtained from different modalities, even
if their value is comparable.
The extent to which beneficiaries are liquidity constrained (i.e., unable to borrow or
convert goods into cash). For example, when food transfers cannot be readily resold, a
“lumpy” cash transfer (unlike a similarly-valued food transfer) would be more likely to
be used to make purchases of larger, nondivisible goods.
Despite substantial research into the impact of food assistance (e.g., Barrett and Maxwell
2005) and the impact of conditional cash transfers (CCTs) in many contexts (see Fiszbein et al.
2009 for review), there is almost no evidence from a rigorous evaluation directly comparing the
impact and cost-effectiveness of cash transfers and food transfers in the same setting (Ahmed et
al. 2009; Gentilini 2007; Webb and Kumar 1995). This evaluation study in Ecuador is one of
several impact evaluations being undertaken in different countries by the World Food
Programme (WFP) and the International Food Policy Research Institute (IFPRI) in which
various forms of transfers will be compared to learn which modalities are most effective in
different contexts.1
1.2 Ecuador Context and Study Objectives
Ecuador has the largest refugee population in Latin America and the vast majority of
refugees are Colombians. In 2010 there were approximately 121,000 refugees and 50,000 asylum
seekers living in Ecuador (UNHCR 2010). As a result of constitutional changes on immigration
and refugee issues in 2008 as well as the Enhanced Registration Program in 2009, Ecuador has
provided documentation to thousands of Colombian refugees in need of international
protection (White 2011). However, despite programmatic action and international support,
refugees remain excluded from many government services and programs, as well as
discriminated against in the job market.
In order to better understand the refugee situation in Ecuador and how to respond to it,
the WFP in collaboration with UNHCR conducted a study on the food security and nutrition
situation of this population. They found that 27.9 percent of the Colombian refugee population
is food insecure and suffers from low dietary diversity (PMA 2010). The incidence of food
insecurity is higher in northern border provinces, reaching 55 percent in Sucumbíos and 44
percent in Carchi and Imbabura. In addition, 48 percent of children are anemic.
Responding to the study’s findings and to a request from the Government of Ecuador
(GoE) in April 2011, WFP expanded its assistance to address the food security needs of refugees
and support their integration into Ecuadorian society. The new program, partially funded by
1 The other countries are Yemen, Niger, and Uganda. These are described, along with the motivation and
learning objectives for this five-country study, in Ahmed et al. (2010).
3
the Spanish Government, used cash, food vouchers, and food transfers to support the most
vulnerable refugees as well as poor Ecuadorians in urban centers of Carchi and Sucumbíos.
Both Carchi and Sucumbíos are northern border provinces that receive high influxes of
Colombian refugees; however, Carchi is located in the northern highlands and Sucumbíos is
located in the Amazonian lowlands.
This evaluation study is targeted at estimating the relative impact and cost-effectiveness
of cash, food vouchers, and food transfers on household food security indicators; as well as
complimentary indicators such as household expenditure and anemia. In addition, this report
investigates the gender effects of the program and the impacts on reducing tensions and
discrimination between Colombian refugees and the host Ecuadorian population. Ecuador is
the only study country testing two alternatives to food assistance (rather than one) as well as
the only study country implementing an urban intervention, thus the potential for learning is
high. In addition to the quantitative impact evaluation, a qualitative study was conducted in
order to learn more about the efficacy of nutrition trainings that accompanied the distribution of
the different transfer types. The study complements the quantitative survey by exploring
beneficiary preference, knowledge retention, and adoption of nutrition content.
This final report describes the context for this study and presents the results. Chapter 2
introduces the WFP food, cash, and voucher transfer program and Chapter 3 describes the
evaluation design. Chapter 4 describes the data and provides summary statistics. Chapter 5
describes the beneficiaries’ experience with the program and nutrition knowledge gained.
Chapter 6 –Chapter 9 presents the results of the impact evaluation on food security, social
capital, anemia, and gender issues. Chapter 10 presents the results from the costing study, and
Chapter 11 introduces the qualitative study on the efficacy of the nutrition training and presents
results. Chapter 12 concludes.
4
2. Intervention
2.1 Site Selection and Beneficiaries
Despite sharing administrative borders, the two intervention provinces of Carchi and
Sucumbíos have markedly different geographic, agro-ecological, and economic characteristics
(Figure 2.1). Carchi is located in the northern highlands at high altitude, characterized by an
industrial and agricultural-based economy including production of tobacco, dairy, floriculture,
and staple crops such as potatoes and maize. Sucumbíos is located in the Amazonian lowlands
and its economy is driven by natural resource extraction, primarily oil, making it one of the
most important economic areas in Ecuador. Sucumbíos is smaller in population as compared to
Carchi (128,995 versus 152,939 inhabitants); however, it is nearly five times the geographical
area (7,076 versus 1,392 square miles) (WFP-Ecuador 2010).
Figure 2.1 Map of intervention provinces
Carchi and Sucumbíos were selected by WFP for the transfer program because of the
high concentration of refugees and poor Ecuadorians as identified by the 2010 Emergency Food
Security Assessment (EFSA). In addition, both provinces have the presence of implementing
partner nongovernmental organizations (NGOs) for food transfers, financial institutions
including ATM branches, supermarkets, and functioning central markets. The urban centers
chosen for the study were Tulcán and San Gabriel in Carchi, and Lago Agrio (also known as
5
Nueva Loja) and Shushufindi in Sucumbíos. Tulcán and Lago Agrio are the capital cities of
Carchi and Sucumbíos, respectively, and represent the largest urban areas within the provinces.
Both cities are in close proximity to the Colombian border (for example, Tulcán is
approximately 7 kilometers from the border). These urban centers were selected by WFP based
on the following criteria:
Total urban population with more than 10 percent refugees;
Poverty index that exceeds 50 percent for the total population;
Presence of implementing partners (NGOs and financial institutions);
In addition to these four primary urban areas, three additional smaller urban areas were
included in the program target population: General Farfan, Huaca, and Pacayacu. These
additional areas are suburbs close to the four primary urban areas, and share the same transfer
distribution centers, financial institutions, and supermarkets.
Barrios (or “neighborhoods”) within these urban centers were then pre-chosen by WFP
in collaboration with local partners as areas that have large numbers of Colombian refugees and
relatively high levels of poverty. Barrios are existing administrative units within the urban
centers and typically headed by Presidents with oversight over social services and other
administrative functions. In November and December of 2010, the Centro de Estudios de Población
y Desarrollo Social (CEPAR) implemented a household census in these barrios. Every household
in the selected barrios was visited, mapped, and administered a one-page questionnaire that
consisted of basic demographic and socioeconomic questions. A proxy means test was then
developed to target beneficiaries based on indicators of household demographics, nationality,
labor force participation, food security, and asset ownership. Based on point scores by
nationality, the decision was made to automatically include all Colombian and Colombian-
Ecuadorian households as “qualifying.” In addition, since WFP wanted to reach households
that were not already being covered by GoE grants, the decision was made to exclude all
households who reported receiving the government’s social safety net transfer program, the
Bono de Desarrollo Humano (BDH). Households residing in the selected barrios with low
socioeconomic status that met the criteria described above formed the original sample eligible
for the intervention.
2.2 Intervention
The intervention consisted of six monthly transfers of food, food vouchers, or cash to
Colombian refugees and poor Ecuadorian households in selected urban centers as previously
described. The transfers were distributed in coordination with local partners and the assistance
of the President of each barrio. An enrollment meeting was held in March before the first
transfer was distributed to issue photo ID cards as well as to sensitize beneficiaries to the
program objectives and logistics. The coordination of the intervention was managed by the
WFP Country Office (CO) through regional sub-offices in each province and was backstopped
by staff at headquarters in Quito. In the initial enrollment and sensitization, the program was
described as a poverty and food security transfer targeted toward women, and therefore, the
majority of the entitlement (program identification) cardholders were expected to be women.
6
However, based on household demographics, men could also be entitlement holders and
participate in all program activities. Overall, approximately 79 percent of cardholders in Carchi
and 73 percent of cardholders in Sucumbíos were women (WFP-Ecuador 2011).
The value of the monthly transfer was standardized across all treatment arms. The total
value of the cash transfer was $40 per month per household. This amount was transferred onto
pre-programmed ATM cards for all cash transfer beneficiaries. Cash transfer households could
retrieve their cash at any time (i.e., they could keep cash in the bank for longer than one month);
however, it had to be taken out in bundles of $10. The food vouchers were also valued at $40
and were given in denominations of $20, redeemable for a list of nutritionally-approved foods
at central supermarkets in each urban center. The list of approved foods is included in
Appendix 1 (along with the recommended amount of food items to buy) and was composed of
cereals, tubers, fruits, vegetables, legumes, meats, fish, milk products, and eggs. The food
vouchers could be used over a series of two visits per month and must be redeemed within 30
days of initial receipt of the voucher. The vouchers were serialized and printed centrally, and
were nontransferable. The food basket was valued according to regional market prices at $40
and included rice (24 kilograms), vegetable oil (4 liters), lentils (8 kilograms), and canned
sardines (8 cans of 0.425 kilograms). The food basket was stored and distributed by local
partners. The transfer size for all modalities was set to be roughly comparable to the national
cash transfer scheme, the BDH, which, at the time of program design, was $35 per household.
Nutrition sensitization was a key component of the program, aimed at influencing
behavior change and increasing knowledge of recipient households, especially in regards to
dietary diversity. To ensure a consistent approach to knowledge transfer, a set of curricula was
developed by WFP to be covered at each monthly distribution and transfers were conditional
on attendance at the nutrition trainings. These topics included (1) program sensitization and
information, (2) family nutrition, (3) food and nutrition for pregnant and lactating women, (4)
nutrition for children aged 0–12 months, and (5) nutrition for children aged 12–24 months.
Nutritional recipes were also distributed throughout the six months. During the last monthly
meeting, an overview and review of lessons learned was implemented, including nutritional
bingo in which participants were asked questions about previous training sessions in game
format. In addition to monthly meetings, posters and flyers were developed and posted at
distribution sites, including supermarkets, banks, and community centers, to further expose
participants to knowledge messaging. These topics included, among others, recommended
consumption of food groups, daily nutritional requirements, proper sanitation, and food
preparation. An example of the posters found outside of supermarkets can be found in
Appendix 1.
From a program design perspective, it is interesting to note the restrictions of specific
transfer modalities. In particular, cash transfers could be categorized as unrestricted (the
household has full control over expenditure decisions), while the voucher could be categorized
as partially restricted (the household must spend the voucher on certain types of food, however,
they have some spending decisions regarding the allocation within the specific food groups),
and the food transfers fully restricted (the household receives a food basket and has no control
over the composition or substitutability). The debate on benefits and drawbacks of cash versus
near-cash versus in-kind benefits is ongoing within different types of transfer programs. On the
one hand, classic economic theory predicts that households are better off and obtain the highest
7
utility from cash transfers which they can allocate towards the domain in which they have the
most need. On the other hand, if certain program objectives are important to achieve, or if
certain characteristics limit participants ability to allocate transfers efficiently (for example
gender or other intra-household issues), then households may benefit from near-cash or in-kind
transfers that are more restricted.
8
3. Evaluation Design
3.1 Study Design
The strategy for estimating the impacts of the cash, food voucher, and food transfer is
built into the design of the program. Sample clusters were randomly assigned to one of four
treatment arms: the cash transfer group, the food voucher transfer group, the food transfer
group, and the comparison group (which received no transfers). Due to the distinct
socioeconomic and geographic characteristics of Sucumbíos and Carchi, the randomization of
cluster centers was stratified at the province level. Stratification guarantees that, within each
stratum, each of the treatment arms is represented equally in each province. The rationale for
doing so is that it prevents the case where, by chance, most clusters assigned to a particular
treatment are located in an area that is very different from another area in which most clusters
are assigned to the other treatment (in this case, location-specific characteristics would be
correlated and confounded with receipt of treatment).
Randomization was conducted in two stages: first, barrios were randomized to either
treatment or comparison arms; second, all treatment clusters (geographical units within barrios)
were randomized to cash, food voucher, or food transfer. This measure was taken to avoid the
case that a cluster assigned to the comparison group might be within the same barrio as a
cluster assigned to receive a transfer and consequently cause discontent by potential
beneficiaries as well as administrators. Conducting the first stage randomization at a higher
level helps minimizing this possibility while still maintaining a sufficient number of
randomization points. The same complications were not foreseen regarding differences in
transfer type across the same barrio, and thus it was possible to randomize the treatment arms
across a smaller disaggregation (clusters).
Before conducting the randomization, consultations were undertaken with WFP and
CEPAR to ensure that (1) there was sufficient number of qualifying households in each cluster
and (2) there was sufficient geographical distinction between clusters. This process led to 63
barrios and 128 clusters in the four urban areas over which to randomize. The number of
clusters per barrio varied from one to six, with an average of approximately two per barrio. The
barrios and clusters were randomized into the four treatment arms using percentages of 20/20
for comparison and food, and 30/30 for cash and food voucher. These percentages were
established in consultation with WFP to meet both beneficiary target sizes by transfer type as
well as sample size requirements for the evaluation design. Approximate sample size
calculations were conducted across countries and can be found in Ahmed et al. (2010).
There are several reasons that we include a comparison group in this study. From an
analytical standpoint, the comparison group allows us to determine whether there are any
impacts of each of the three treatments rather than merely if there are any differences in the
impacts between the three treatments. For example, without a comparison group, we would be
unable to distinguish the case where food, voucher, and cash groups both have identical large
positive impacts on our outcomes of interest from the case where food, voucher, and cash all
have no impact. Moreover, the presence of a comparison group allows us to disentangle
impacts of the treatment from other time trends. In particular, if both the treated groups look
worse off on a range of outcomes at endline and there were no comparison group, it would be
9
impossible to determine whether receiving transfers somehow caused households to be worse
off or whether secular trends over time that affected all households regardless of transfer receipt
(e.g, weather conditions, market conditions) were responsible for the decline. The presence of
the comparison group allows us to difference out the outcomes that we would counterfactually
expect for the treatment groups in the absence of being treated, and therefore to isolate the
change in outcomes that can be attributed exclusively to the treatments. From the practical
standpoint, given that there are program funding ceilings and transfers cannot be given to all
potentially vulnerable households, it is also a fair method to randomly assign which clusters
and corresponding households receive transfers and which do not.
One unexpected complication in the study design was the change in beneficiary criteria
implemented during the baseline survey data collection. In the process of surveying
households, it was concluded that the targeting for the transfers was too broad, resulting in the
inclusion of households who were relatively well off as compared to the program objectives.
This led to a re-targeting process where households who were relatively well off were dropped
from the study. Since there were not enough households in existing barrios to replace those that
had been excluded, the decision was made to expand coverage to additional barrios on the
outer circle of urban areas. These areas were subsequently re-randomized into treatment arms
according to the approximate percentage lost in each treatment arm. This included ten new
clusters in Sucumbíos and eight new clusters in Carchi.
3.2 Study Sample
In order to conduct an evaluation of the cash, voucher, and food transfer program,
baseline and follow-up surveys were conducted. The baseline survey was conducted in March-
April 2011 before the first transfers were distributed. The follow-up survey was conducted
approximately 6 months later (October-November 2011) after the last of the six distributions.
The sampling for the baseline survey was conducted by CEPAR and IFPRI after
receiving the beneficiary lists. Based on the distribution of clusters in the treatment arms and
the required sample sizes, 27 households per comparison and food clusters and 20 households
per food voucher and cash clusters were randomly selected to be interviewed in the baseline
survey. In addition, since one of the main objectives of the evaluation was to compare
differences across nationalities, the Colombian and Colombian-Ecuadorian households were
oversampled to ensure a sufficient sample for comparative analysis. In total, 3,331 households
were surveyed at baseline. However, approximately 30 percent were too rich and subsequently
excluded from the program, and so we focus only on 2,357 households that were included after
the retargeting period and could be matched with monitoring data. Of these households, 2,122
were re-surveyed at follow-up.
3.3 Estimation Strategy
Our estimation strategy relies on the randomized design of the cash, voucher, and food
transfer program. Because the total number of clusters is relatively large, random assignment of
clusters assures that, on average, households will have similar baseline characteristics across
treatment and comparison arms. Such a design eliminates systematic differences between
10
beneficiaries and non-beneficiaries in targeted programs and minimizes the risk of bias in the
impact estimates due to “selection effects” based on differences in household characteristics. As
a result, average differences in outcomes across the groups after the intervention can be
interpreted as being truly caused by, rather than simply correlated with, the receipt of transfers.
Moreover, we take advantage of the baseline survey and estimate the treatment effect
using Analysis of Covariance (ANCOVA) which controls for the lagged outcome variable.
Given the high variability and low autocorrelation of the data at baseline and follow-up,
ANCOVA estimates are preferred over difference-in-difference estimates (McKenzie 2012).
Intuitively, if autocorrelation is low, then difference-in-difference estimates will overcorrect for
baseline imbalances. ANCOVA estimates on the other hand will adjust for baseline imbalances
according to the degree of correlation between baseline and follow-up. The ANCOVA model
that we estimate is the following:
where is the outcome of interest for household at follow-up and is the outcome of
interest at baseline. is an indicator for whether household is in the treatment group, and is
the ANCOVA estimator. In other words, represents the amount of change in outcome, Y,
which is due to being assigned to the treatment group. For example, if the outcome Y is
monthly household per-capita food expenditure measured in USD, and is 3.41 and significant,
the interpretation is that being in the treatment group leads to on average, a 3.41 USD increase
in per capita food expenditure over the intervention period, holding other covariates constant.
X is a vector of baseline covariates and is a vector of coefficients that correspond to each
covariate. For each unit change in a specific variable x the corresponding represents the
amount of change in outcome Y that is due to x holding other covariates constant.
Given the relative success of the random assignment, the inclusion of baseline controls is
not necessary to obtain unbiased estimates of . In most estimates, however, we account for
socioeconomic characteristics from the baseline survey in order to increase the precision of the
estimates and control for any minor differences between treatment and comparison arms at
baseline. The core group of baseline “control” variables that we use are indicators for urban
centers, an indicator for whether household head is female, and indicator for whether
household head is Colombian, an indicator for whether household head has secondary
education or higher, household head’s age, household size, number of children 0–5 years old,
number of children 6–15 years old, and household wealth quintiles (five indicators for each
quintile). The household wealth quintiles were constructed from a wealth index that was
created using the first principal from a principal components analysis (PCA), which is similar to
the methodology used in the Demographic and Health Survey (DHS) to construct wealth scores.
Variables used to construct the index are housing infrastructure indicators (e.g., type of floor,
roof, toilet, light, fuel, and water source) and 11 asset indicators (e.g., refrigerator, mobile
phone, TV, car, and computer). In all regressions we cluster the standard errors at the level of
randomization that is the cluster.
The regression equation above can be expanded to estimate the ANCOVA estimator for
each treatment arm. Specifically,
11
where is an indicator for whether household is in the food treatment arm, is an
indicator for whether the household is in the cash treatment arm, and is an indicator
for whether the household is in the voucher treatment arm. The corresponding to each
treatment arm represents the ANCOVA estimator for each arm. In order to test whether the
ANCOVA estimator is statistically different by treatment arm, we conduct Wald tests of
equality.
Given that we are interested in whether the impact of treatment differed by whether or
not the household is Colombian, we also estimate the differential effect of treatment by creating
an interaction term of the treatment indicator (T) with the indicator for whether or not the
household head is Colombian (C). Specifically, we estimate
now corresponds to the impact of being in the treatment arm for Ecuadorian households,
while corresponds to the impact of being in the treatment arm for Colombian
households. Thus, is the differential impact with respect to being Colombian of the pooled
treatment. Similar interaction terms are created across treatment arms to estimate the
differential effect with respect to being Colombian for each treatment arm separately.
12
4. Data
4.1 Survey Instruments and Topics
The survey instruments used for the baseline and follow-up consisted of several
components:
Household questionnaire: completed for each household in the sample;
Hemoglobin questionnaire: completed for children aged 6—59 months at baseline; and
adolescent girls aged 10—16 years in each designated household;
Barrio questionnaire: completed for each barrio;
Central market and supermarket price questionnaire: completed for each urban center.
An important feature of the survey instruments was that the structure and content of the
questionnaires remained largely unchanged across survey rounds. This comparability means
that interpreting changes in outcomes over time is not confounded by changes in the questions
used to elicit these data.
The household questionnaire contains household-level information as well as detailed
information on individual household members. As the key objective of the study is to
understand how households use the transfers, whether the use differs by transfer modality, and
which household and environmental characteristics determine use, many of the modules in the
household questionnaire focus on household socioeconomic characteristics and uses/sources of
resources. Moreover, as an objective of the study is to understand the interplay between transfer
receipt, gender, and nationality, several modules in the household questionnaire are devoted to
capturing intra-household allocation of resources, decisionmaking within the household, and
discrimination. The exact modules used at baseline can be found in Table 4.1. Most sections in
the household questionnaire are answered by the member of the household who is most
knowledgeable on the topic, which in the majority of cases is either the household head or
spouse. In addition to the modules at baseline, the follow-up questionnaire contains a module
on households’ experience with the cash, voucher, and food transfer program.
In addition to the household questionnaire, a one-page hemoglobin questionnaire was
completed for children 6–59 months at baseline and adolescent females age 10–16 years old. The
questionnaire collected basic information, including detailed birth information, and health
information relevant for calculating anemia measurements such as pregnancy and menstruation
status (for adolescent girls). In addition, the hemoglobin questionnaire was designed to allow
for a rapid field adjustment for altitude and pregnancy status, to properly refer children and
girls who were found to be anemic, or below the cut-off 12 g/dl (adolescent girls) and 11 g/dl for
children aged 6 months to 5 years, to local health clinics. All enumerators carried referral cards
to local health clinics during fieldwork so that appropriate information and referrals could be
made as fieldwork was conducted.
13
Table 4.1 Modules in the baseline household questionnaire
Module Description Target respondent
A Household identification, location, and interview details Household head
B Household roster and demographic information Household head
C Education (all members aged 4–18) Household head
D Activities and labor force participation (all members aged 12 and above) Household head
E Dwelling characteristics Household head
F Health Female head/spouse
G Maternal health Female head/spouse
H Health and nutrition knowledge Female head/spouse
I Consumption habits and food security indicators Female head/spouse
J Consumption and food expenditure Female head/spouse
K Food frequency (children 6–59 months and adolescent girls aged 10–16) years Female head/spouse
L Markets and expenditure behavior Female head/spouse
M Nonfood expenditure Household head
N Assets (productive, durable, and credit) Household head
O Other transfers and income sources Household head
Q Budgeting behavior Household head
R Perceptions and discrimination Household head
S Migration Household head
T Women’s status, decisionmaking, and domestic violence Female head/spouse
Barrio questionnaires were completed for each barrio included in the sample. The
instrument was administered to the barrio president, administrators, or “key informants” in the
neighborhoods. The questionnaires included information on community characteristics;
educational and health facilities; access to services in the community; infrastructure; livelihoods
and shocks; and women’s status in the community. The final component of the barrio
questionnaire was a food price survey. The food price survey included a list of the main food
items from the household food consumption module, to determine if food was available in
barrio markets or shops, and if so, asking the unit price of the food item. This information
reveals to what extent variation in availability and price of food items explains the variation in
consumption patterns.
Finally, a central market and supermarket price questionnaire was collected, which consisted
of the same price listings that were included as part of the barrio questionnaire. These
questionnaires were administered in the participating supermarket and the city central market
in each urban area. The objective of this exercise is to allow us to analyze the cost differentials
faced by households using food vouchers, in contrast to those receiving food or cash.
4.2 Attrition Analysis
As mentioned in the section above, our analysis focuses on 2,357 households at baseline,
of which we were able to resurvey 2,122, yielding an overall attrition rate of 10 percent (Table
4.2). In Carchi, attrition rates are slightly higher than attrition rates in Sucumbíos, 10.7 percent
14
versus 9.5 percent. A 10 percent attrition rate across survey rounds is comparable to the internal
overall program attrition rate reported by WFP of approximately 8 percent over five cycles
(WFP-Ecuador 2011). One reason for the fairly high attrition rate over such a short period of
time is that households in our sample live near the Colombian border and are thus fairly
transient. In many cases the border area is geographically adjacent to the surveyed barrios and
households engage in cross-border activities such as trade and movement of goods and
services. In order to find jobs, households also move to larger urban centers such as Quito and
Guayaquil. The protocol for tracking households was as follows: if households lived in the same
urban center however changed clusters, they were tracked with multiple visits attempted at
different times during the fieldwork period. In addition, if households moved to another of the
eight survey urban centers, or to Quito, they were tracked. Households that moved to a rural
area or province outside the study area were not tracked.
Table 4.2 Attrition rates
Households at baseline Households at follow-up Attrition rates
All 2357 2122 9.97
By province
Carchi 916 818 10.70
Sucumbíos 1441 1304 9.51
For the purposes of our impact evaluation, if attrition is correlated with treatment
assignment, then this could potentially bias our impact estimates. Table 4.3 shows that there is
not a significant difference in attrition rates between the comparison arm and the pooled
treatment arm. However, when we compare attrition across the comparison arm and each
individual treatment arm, we do find a significant difference in rates for the food transfer arm
when compared to the comparison arm (Table 4.4).
Table 4.3 Attrition rates, by treatment and comparison groups
Comparison Treatment Difference
Attrition rates 0.11 0.09 0.02
(0.01) (0.01) (0.01)
N 652 1,705
Notes: Treatment refers to all treatment arms (food, cash, and voucher) combined. Standard errors
reported in parentheses. Difference in means conducted using t-tests. Stars indicate the following
significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01.
Table 4.4 Difference in attrition rates, by treatment arms
Means Difference in means
Comparison Food Cash Voucher
Comparison-
food
Comparison-
cash
Comparison-
voucher
Attrition rates 0.11 0.08 0.09 0.11
0.04** 0.02 0.01
(0.01) (0.01) (0.01) (0.01)
(0.02) (0.02) (0.02)
N 652 453 601 651
Notes: Standard errors reported in parentheses. Difference in means conducted using t-tests. Stars indicate the
following significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01.
15
Attrition could bias the study results in a number of unanticipated ways. If poorer
households are the ones more likely to leave the study, and significantly more households in
the comparison arm left the study than in the food treatment arm, then our estimates will be
underestimates of the impact of food transfers. Even across arms with similar attrition rates,
differential attrition could threaten the internal validity of the study. In particular, if households
that leave the treatment arms are poorer than households that leave the comparison arm, then
our treatment estimates will be biased because any change in outcomes will be due to both
treatment and differential attrition. In order to examine if differential attrition threatens the
internal validity of the study, we compare baseline characteristics of households that leave the
study across treatment and comparison arms. Table 4.5 reveals that in the treatment and
comparison arms, there are significant differences between those that leave the study and those
that stayed. For example, in both the treatment and comparison arms, Colombians are
significantly more likely to leave the study. However, internal validity is only threatened if
those that leave the study in the comparison arm are significantly different from those that leave
the treatment arm. Consequently, we focus on columns 7 and 8 that compare the “attrited”
groups across the two arms, and we find significant differences at the 5 percent level only for
household head’s age and ownership of agricultural plots. In particular, those that leave the
treatment arm are significantly younger and less likely to own agricultural land than those that
leave the comparison arm. However, baseline analysis across treatment and comparison arms
for households that are left in our study (Table 4.6) reveals that differences in age and owning
agricultural land are not significant; therefore, we conclude that the bias due to the differential
attrition of these variables is likely to be very small.
4.3 Baseline Analysis (by Treated and Comparison)
In order to ensure that randomization was successful, we compare baseline
characteristics across treatment and comparison households. We conduct the analysis on the
2,122 households that are in the baseline and follow-up surveys, however, a complete analysis
for all baseline households can be found in the baseline technical report (Hidrobo et al. 2011).
We first combine all three treatment arms (cash, voucher, and food transfer) and compare
pooled treatment households to comparison households, and then we compare each treatment
arm separately to the comparison arm.
16
Table 4.5 Differential attrition analysis (mean values of baseline characteristics)
Comparison arm Treatment Arm Difference
(1) (2) (3) (4) (5) (6) (7) (8)
Attrited In study P-value Attrited In study P-value Col(1)-Col(4) P-value
Characteristics of the household head
Male 0.69 0.74 0.36 0.66 0.73 0.10 0.02 0.71
Age 42.08 41.87 0.90 36.67 41.63 0.00 5.41 0.01
Has primary education 0.53 0.60 0.21 0.48 0.58 0.03 0.04 0.55
Has secondary education or higher 0.41 0.32 0.14 0.47 0.36 0.01 -0.06 0.39
Married 0.23 0.28 0.41 0.22 0.28 0.10 0.01 0.83
Colombian 0.57 0.37 0.00 0.55 0.26 0.00 0.01 0.83
Household characteristics
Floor type: dirt 0.08 0.06 0.37 0.03 0.04 0.65 0.05 0.09
Cooking fuel: gas 0.89 0.96 0.01 0.89 0.95 0.00 -0.00 0.95
Drinking water: tap 0.59 0.49 0.08 0.57 0.46 0.01 0.03 0.67
Waste system: private toilet 0.46 0.58 0.04 0.37 0.48 0.01 0.09 0.21
Household size 3.66 4.01 0.17 3.34 3.75 0.01 0.33 0.19
Number of children 0–5 years 0.57 0.58 0.92 0.59 0.62 0.65 -0.02 0.83
Number of children 6–15 years 0.80 1.01 0.15 0.72 0.86 0.11 0.08 0.58
Household property (percent who own asset)
Mobile telephone 0.82 0.84 0.69 0.81 0.82 0.81 0.01 0.85
TV and/or DVD 0.69 0.80 0.03 0.64 0.81 0.00 0.05 0.46
Washing machine 0.28 0.41 0.03 0.25 0.34 0.02 0.04 0.57
Car/truck/motorcycle 0.20 0.24 0.51 0.17 0.23 0.08 0.03 0.60
Agricultural plot 0.12 0.13 0.91 0.05 0.13 0.00 0.07 0.05
Monthly household expenditure 250.57 250.75 0.99 249.83 256.47 0.72 0.73 0.98
Notes: In columns (3) and (6), p-values are reported from t-tests on the equality of means for each variable between the “In Study” and “Attrited” groups. Column
(7) reports the difference in means between the “Attrited” group in the comparison arm and the “Attrited” group in the treatment arm. Column (8) reports the p-
values for the difference in means between the two “Attrited” groups. “In study” sample consists of households that were in the baseline and follow-up.
“Attrited” refers to households that were in the baseline survey but not in the follow-up.
17
Table 4.6 Baseline characteristics, by treatment and comparison arms
Variable All Comparison Treatment Difference
Characteristics of the household head
Male 0.73 0.74 0.73 -0.01
(0.01) (0.02) (0.01) (0.02)
Age 41.69 41.87 41.63 -0.24
(0.33) (0.61) (0.40) (0.73)
Has primary education 0.58 0.60 0.58 -0.03
(0.01) (0.02) (0.01) (0.02)
Has secondary education or higher 0.35 0.32 0.36 0.04
(0.01) (0.02) (0.01) (0.02)*
Married 0.28 0.28 0.28 0.00
(0.01) (0.02) (0.01) (0.02)
Colombian 0.29 0.37 0.26 -0.11***
(0.01) (0.02) (0.01) (0.02)
Household characteristics
Floor type: dirt 0.04 0.06 0.04 -0.02
(0.00) (0.01) (0.00) (0.01)
Cooking fuel: gas 0.96 0.96 0.95 0.00
(0.00) (0.01) (0.01) (0.01)
Drinking water: tap water 0.47 0.49 0.46 -0.02
(0.01) (0.02) (0.01) (0.02)
Waste system: private toilet 0.51 0.58 0.48 -0.10***
(0.01) (0.02) (0.01) (0.02)
Household size 3.82 4.01 3.75 -0.26***
(0.04) (0.09) (0.05) (0.10)
Number of children 0-5 years 0.61 0.58 0.62 0.04
(0.02) (0.03) (0.02) (0.04)
Number of children 6–15 years 0.90 1.01 0.86 -0.14**
(0.02) (0.05) (0.03) (0.06)
Family property (percent ownership)
Mobile telephone 0.83 0.84 0.82 -0.02
(0.01) (0.02) (0.01) (0.02)
TV 0.80 0.80 0.81 0.00
(0.01) (0.02) (0.01) (0.02)
Washing machine 0.36 0.41 0.34 -0.07***
(0.01) (0.02) (0.01) (0.02)
Car / truck / motorcycle 0.23 0.24 0.23 0.00
(0.01) (0.02) (0.01) (0.02)
Agricultural plot 0.13 0.13 0.13 0.00
(0.01) (0.01) (0.01) (0.02)
Monthly household expenditure 254.91 250.75 256.47 5.71
(4.85) (8.66) (5.83) (10.44)
Number of observations 2,122 578 1,544
Notes: Treatment refers to all treatment arms (food, cash, and voucher) combined. Standard errors reported in
parentheses. Difference in means conducted using t-tests. Stars indicate the following significance levels: * p < 0.10, **
p < 0.05, *** p < 0.01.
18
Table 4.6 reveals that household heads in our sample have a mean age of 41.69 years ,
most have only primary education (58 percent), only 28 percent are married (while 38 percent
are in a civil union or co-habiting), 29 percent are Colombian, and nearly three-quarters are
males (73 percent). Approximately 96 percent of households use gas for cooking and
approximately half have a private toilet. The average household size is 3.8 and most households
have cellular phones (83 percent) and televisions/DVDs (80 percent). Average monthly
household expenditure is approximately $255, of which 46.4 percent is allocated to food
purchases, 38.8 percent to nonfood purchases, 5.8 percent to education-related expenses, and 10
percent to health-related expenses (Figure 4.1). As discussed in the baseline report, the statistics
on household size are not surprising, given the sampling procedure and inclusion criteria for
the interventions. In particular, we expect these households to be relatively small, because they
are households that do not qualify for the BDH, and thus are less likely to have young or school
age children.
Figure 4.1 Composition (by shares) of monthly household expenditure
Across 19 difference-in-means test between the treatment group and comparison group,
shown in the last column of Table 4.6, only 5 are statistically different at the 5 percent level,
which reveals that randomization was, for the most part, effective at balancing baseline
characteristics. In particular, comparison households are significantly more likely to be
Colombian, have more children ages 6–15 years, have larger households, have a private latrine,
and have a washing machine.
Table 4.7 is similar to Table 4.6; however, it conducts difference-in-means test for each
treatment arm compared to the comparison arm. Results show a similar pattern where, across
the 57 (19 x 3) tests, 13 have means that are significantly different at the 5 percent level (these
are the same variables that are statistically different in the pooled treatment table, with the
addition of the variables on floor type and household head’s secondary education). Overall,
these confirm previous tests by pooled treatments that indicate that the baseline randomization
5.8 6.1 5.6
10.0 9.7 10.1
38.8 39.6 38.5
46.4 46.0 46.5
0%
20%
40%
60%
80%
100%
All Comparison Treatement
Food
Non-food items
Health
Education
N=2,122
19
was generally successful with respect to household observable characteristics. However, the few
significant differences reaffirm our decision to add baseline covariates as controls in our
empirical analysis.
Table 4.7 Baseline characteristic, by treatment arm
Variable
Means
Difference in Means
Comparison Food Cash Voucher
Food -
Comparison
Cash -
Comparison
Voucher -
Comparison
Characteristics of the head
Male 0.74 0.75 0.72 0.71
0.01 -0.02 -0.02
(0.02) (0.02) (0.02) (0.02)
(0.03) (0.03) (0.03)
Age 41.87 41.26 41.47 42.04
-0.61 -0.39 0.17
(0.61) (0.75) (0.66) (0.66)
(0.97) (0.90) (0.90)
Has primary education 0.60 0.58 0.58 0.57
-0.02 -0.03 -0.03
(0.02) (0.02) (0.02) (0.02)
(0.03) (0.03) (0.03)
Has secondary education or higher 0.32 0.35 0.35 0.38
0.03 0.03 0.06
(0.02) (0.02) (0.02) (0.02)
(0.03) (0.03) (0.03)**
Married 0.28 0.30 0.28 0.26
0.03 0.00 -0.01
(0.02) (0.02) (0.02) (0.02)
(0.03) (0.03) (0.03)
Colombian 0.37 0.28 0.25 0.27
-0.10*** -0.13*** -0.11***
(0.02) (0.02) (0.02) (0.02)
(0.03) (0.03) (0.03)
Household characteristics
Floor type: dirt 0.06 0.04 0.03 0.04
-0.01 -0.02** -0.01
(0.01) (0.01) (0.01) (0.01)
(0.01) (0.01) (0.01)
Cooking fuel: gas 0.96 0.96 0.95 0.96
0.00 -0.01 0.00
(0.01) (0.01) (0.01) (0.01)
(0.01) (0.01) (0.01)
Drinking water: tap water 0.49 0.44 0.51 0.44
-0.05 0.03 -0.05*
(0.02) (0.02) (0.02) (0.02)
(0.03) (0.03) (0.03)
Waste system: private toilet 0.58 0.52 0.46 0.49
-0.07** -0.13*** -0.09***
(0.02) (0.02) (0.02) (0.02)
(0.03) (0.03) (0.03)
Household size 4.01 3.82 3.75 3.69
-0.18 -0.26** -0.32***
(0.09) (0.09) (0.08) (0.07)
(0.12) (0.11) (0.11)
Number of children 0-5 years 0.58 0.66 0.59 0.62
0.08 0.01 0.04
(0.03) (0.04) (0.03) (0.03)
(0.05) (0.05) (0.05)
Number of children 6-15 years 1.01 0.89 0.88 0.82
-0.11 -0.13* -0.18***
(0.05) (0.05) (0.05) (0.04)
(0.07) (0.07) (0.07)
Family property
Mobile telephone 0.84 0.81 0.82 0.83
-0.04 -0.02 -0.01
(0.02) (0.02) (0.02) (0.02)
(0.02) (0.02) (0.02)
TV 0.80 0.81 0.78 0.82
0.01 -0.02 0.02
(0.02) (0.02) (0.02) (0.02)
(0.03) (0.02) (0.02)
Washing machine 0.41 0.36 0.31 0.34
-0.05* -0.10*** -0.07**
(0.02) (0.02) (0.02) (0.02)
(0.03) (0.03) (0.03)
Car / truck / motorcycle 0.24 0.22 0.23 0.25
-0.02 0.00 0.01
(0.02) (0.02) (0.02) (0.02)
(0.03) (0.03) (0.03)
Own agricultural plot 0.13 0.12 0.12 0.13
-0.01 -0.01 0.01
(0.01) (0.02) (0.01) (0.01)
(0.02) (0.02) (0.02)
Monthly household expenditure 250.75 284.22 249.16 243.35 33.47* -1.59 -7.40
(8.66) (14.88) (9.19) (7.09) (17.22) (12.62) (11.19)
N 578 418 545 581
Notes: Standard errors reported in parentheses. Difference in means conducted using t-tests. Stars indicate the
following significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01.
20
5. Experience with Program
5.1 Experience with Program
In this section, we present indicators describing beneficiaries’ experiences with the
program. The tables presented in this section are descriptive and intended to provide insight on
beneficiaries’ costs, preferences and use of the transfer. The aim is to provide context for
potential pathways through which the transfers work to achieve food security and improve
social capital and gender impacts, and to be able to compare across modalities so that we can
better understand the impact results presented in Chapters 6 through 9. It should be noted that
these statistics are self-reported by beneficiaries when asked directly about the transfer and are
not comparable to the impact evaluation measures used in Chapter 6 which are constructed
from consumption modules and household aggregates and thus are more reliable, objective
measures. In addition, the sample for this analysis excludes the comparison group since they
did not participate in the program. Moreover, due to attrition in program enrollment and over
the course of the program, not everyone who was in a treatment cluster participated in the
program, and thus the sample in this section is smaller than the sample we use in the rest of our
analysis.
We start by describing total costs to beneficiaries of participation in the transfer
program, both in terms of time and out of pocket expenditures (Table 5.1). On average,
individuals took 31 minutes to travel from their homes to distribution points, spent 44 minutes
waiting at distribution points, and spent $1.72 in total to receive the transfer (equivalent to 4.3
percent of the monetized transfer value). Beneficiaries from Sucumbíos had significantly higher
averages across all three costs, with, on average, 8 minutes longer travel time to the distribution
points, 23 minutes longer wait times and spent an average of $1.25 more in transport costs as
compared to Carchi. These differences could be explained by a number of factors, including the
larger geographic spread of beneficiaries in Sucumbíos, as well as potentially larger numbers of
beneficiaries assigned to each distribution point.
Table 5.1 Time and cost to receive the transfer, by province
The last time you have received the food/cash/voucher transfers: All Carchi Sucumbíos Difference
How long did it take you to get to the distribution point, from the
place that you live? 31.33 26.93 34.46 -7.53***
(0.75) (0.81) (1.13) (1.39)
How long did you wait to receive your transfer, from the moment
you arrived at the distribution point? 44.00 30.66 53.50 -22.85***
(1.69) (1.73) (2.57) (3.10)
How much did it cost you (total) to receive the transfer
(transportation, etc.) 1.72 0.99 2.23 -1.25***
(0.15) (0.10) (0.24) (0.26)
Number of observations 1,207 502 705
Notes: Mean values reported with standard errors in parentheses below means. * indicates significance at the 10
percent level, ** significance at the 5 percent level, and *** significance at the 1 percent level.
When comparing across the different treatment modalities (Table 5.2), we see that for
food beneficiaries, the amount of time to reach the food distributions points is significantly
longer (approximately 39 minutes versus 28 minutes) as compared to cash and voucher
beneficiaries. However, the average time spent waiting to receive vouchers and food was much
21
higher than that for cash (approximately 63 and 54 minutes for voucher and food households in
comparison to 16 minutes for cash households). This difference is not surprising since cash
beneficiaries could withdrawal their money at ATM machines rather than cueing at a specified
time like the voucher and food households. However, the positive waiting time even for cash
recipients reflects extra assistance needed for beneficiaries to withdraw money and availability
of ATM machines.
Table 5.2 Time and cost to receive transfers, by modality
The last time you have received the
food/cash/voucher transfers: Food Cash Voucher
Food -
Cash
Food -
Voucher
Cash -
Voucher
How long did it take you to get to the
distribution point, from the place that you live?
38.87 28.39 28.34 10.48*** 10.53*** 0.04
(1.67) (1.19) (1.02) (2.05) (1.96) (1.57)
How long did you wait to receive your transfer,
from the moment you arrived at the distribution
point?
53.92 16.16 63.16 37.76*** -9.23* -46.99***
(3.56) (1.00) (3.21) (3.70) (4.80) (3.36)
How much did it cost you (total) to receive the
transfer (transportation, etc.)
2.12 1.46 1.65 0.66** 0.46* -0.20
(0.10) (0.31) (0.26) (0.32) (0.28) (0.40)
Number of observations 341 425 441
Notes: Mean values reported with standard errors in parentheses below means. * indicates significance at the 10
percent level, ** significance at the 5 percent level, and *** significance at the 1 percent level.
Table 5.3 presents descriptive statistics on the transparency and administrative processes
of the program and shows that, in general, the functioning was very successful. For example, on
average, 97 percent of the beneficiaries report receiving sufficient information needed to
understand how the program works. On average, 99 percent of beneficiaries report receiving
their transfer in totality, believe the program is fair, and report program employees treated
them with respect. In addition, 98 percent of beneficiaries trust their transfers will not be stolen.
Slightly lower percentages of beneficiaries report that they received transfers at the scheduled
times (approximately 91 percent) and that they knew how many more transfers they would
receive (approximately 93 percent). There are few significant differences by province in these
indicators, with beneficiaries in Sucumbíos reporting slightly lower overall understanding of
how the program works (95 percent in comparison to 98 percent) and slightly lower reports of
receiving the transfer on time and being treated with respect.
Table 5.4 replicates these results stratifying on treatment modality. Results show that the
cash arm has a lower percentage of respondents reporting receiving the transfer on time
compared to the food and voucher arms; and the food arm has the highest percentage of
respondents knowing the number of transfers they would be receiving.
22
Table 5.3 Transparency and administrative processes, by province
Statements (true = 1, false = 0) All Carchi Sucumbíos Difference
In general, you have received all information needed to
understand how the program works
0.97 0.98 0.95 0.03***
(0.01) (0.01) (0.01) (0.01)
In general, you received the transfers at the scheduled time 0.91 0.93 0.90 0.03*
(0.01) (0.01) (0.01) (0.02)
In general, you received the transfers in its totality 0.99 0.99 0.98 0.01
(0.00) (0.00) (0.00) (0.01)
In general, the program employees treated you with respect 0.99 1.00 0.99 0.01*
(0.00) (0.00) (0.00) (0.01)
In general, you knew how many transfers you would
receive in the future
0.93 0.94 0.92 0.02
(0.01) (0.01) (0.01) (0.02)
In general, you trust the program is fair and will help your
family
0.99 1.00 0.99 0.01
(0.00) (0.00) (0.00) (0.00)
In general, you are confident your food/money/voucher is
safe and will not be stolen
0.98 0.98 0.98 0.00
(0.00) (0.01) (0.01) (0.01)
Participating at the program has helped you meet and talk
to people you would normally not have known
0.99 0.99 0.98 0.01
(0.00) (0.00) (0.00) (0.01)
Number of observations 1,207 502 705
Notes: Mean values reported with standard errors in parentheses below means. * indicates significance at the 10
percent level, ** significance at the 5 percent level, and *** significance at the 1 percent level.
Table 5.4 Transparency and administrative processes, by modality
Statements (true =1, false=0) Food Cash Voucher
Food -
Cash
Food -
Voucher
Cash -
Voucher
In general, you have received all information
needed to understand how the program works
0.97 0.97 0.96 0.00 0.01 0.01
(0.01) (0.01) (0.01) (0.01) (0.01) (0.01)
In general, you received the transfers at the
scheduled time
0.94 0.88 0.93 0.06*** 0.01 -0.05**
(0.01) (0.02) (0.01) (0.02) (0.02) (0.02)
In general, you received the transfers in its totality 0.99 0.98 0.99 0.00 -0.01 -0.01
(0.01) (0.01) (0.00) (0.01) (0.01) (0.01)
In general, the program employees treated you
with respect
0.99 0.99 0.99 0.00 0.01 0.00
(0.00) (0.00) (0.01) (0.01) (0.01) (0.01)
In general, you knew how many transfers you
would receive in the future
0.97 0.93 0.88 0.03** 0.08*** 0.05**
(0.01) (0.01) (0.02) (0.02) (0.02) (0.02)
In general, you trust the program is fair and will
help your family
1.00 0.99 0.99 0.00 0.00 0.00
(0.00) (0.00) (0.00) (0.01) (0.00) (0.01)
In general, you are confident your
food/money/voucher is safe and will not be
stolen
0.98 0.98 0.98 0.00 0.01 0.00
(0.01) (0.01) (0.01) (0.01) (0.01) (0.01)
Participating at the program has helped you meet
and talk to people you would normally not have
known
0.99 0.99 0.98 -0.01 0.00 0.01
(0.01) (0.00) (0.01) (0.01) (0.01) (0.01)
Number of observations 341 425 441
Notes: Mean values reported with standard errors in parentheses below means. * indicates significance at the 10
percent level, ** significance at the 5 percent level, and *** significance at the 1 percent level.
To elicit preferences around transfer modalities, beneficiaries were asked how they
would have liked to receive their transfers, in cash, voucher, or food. In general, a higher
23
number of respondents prefer receiving all the transfer in cash; however, the response is very
influenced by what participants are actually receiving (food, cash, or voucher). In particular, the
majority of individuals report preferring the way they currently receive transfers, with 55
percent of individuals in the food arm, 77 percent of individuals in the cash arm, and 56 percent
of individuals in the voucher arm, choosing to receive 100 percent of their transfers in food,
cash, and voucher, respectively (Table 5.5). This result, in part, signifies that beneficiaries are
satisfied with the status quo, which could reflect the satisfaction of receiving something “free”
in general. On the other hand, a sizable percentage of individuals report preferring to receive
none of the transfer that they are currently receiving. In particular, out of the total surveyed
voucher beneficiaries, 31 percent prefer not to receive any of their transfer in voucher, while for
food beneficiaries, 28 percent prefer not to receive any in food and only 9 percent of cash
beneficiaries prefer not to receive any in cash. Thus, it appears that, on average, cash
households are the most satisfied with their transfer modality, followed by food and then
voucher households.
Table 5.5 Satisfaction with transfer modality, by treatment status
How would you like to receive your transfer? All Food Cash Voucher
All in cash 0.37 0.07 0.77 0.20
(0.01) (0.01) (0.02) (0.02)
All in voucher 0.26 0.18 0.02 0.56
(0.01) (0.02) (0.01) (0.02)
All in food 0.21 0.55 0.07 0.08
(0.01) (0.03) (0.01) (0.01)
None in cash 0.49 0.77 0.09 0.66
(0.01) (0.02) (0.01) (0.02)
None in voucher 0.65 0.75 0.92 0.31
(0.01) (0.02) (0.01) (0.02)
None in food 0.68 0.28 0.83 0.86
(0.01) (0.02) (0.02) (0.02)
Number of observations 1,207 341 425 441
Note: Mean values reported with standard errors in parentheses.
The results in Table 5.5 can be in part explained by the reported difficulties experienced
by beneficiaries, which were more commonly experienced by voucher beneficiaries. In
particular, 79 percent of voucher beneficiaries report at least one complaint with their transfer,
compared to 40 percent in the cash group and 37 percent in the food group. Table 5.6 reveals the
numerous difficulties that voucher beneficiaries report, including high prices in supermarkets
(66 percent), lack of food in supermarkets (48 percent), problems with payment (11 percent),
and lack of understanding on how to use the voucher (10 percent). Beneficiaries in Sucumbíos
were significantly less likely to cite high prices (59 percent as compared to 78 percent), but were
significantly more likely to report dislike of the purchasable voucher foods (8 percent as
compared to 3 percent). Approximately 10 percent of the beneficiaries noted “other” problems,
which could be, among other things, restrictions on the days of the week or month that
beneficiaries could utilize vouchers.
24
Table 5.6 Difficulties experienced by voucher beneficiaries, by province
Indicator (=1) if true All Carchi Sucumbíos Difference
Lack of food in supermarkets 0.48 0.46 0.49 -0.03
(0.02) (0.04) (0.03) (0.05)
Problems in the cashier (with payment, etc.) 0.11 0.10 0.11 -0.02
(0.01) (0.02) (0.02) (0.03)
Did not understand how to use the vouchers 0.10 0.09 0.11 -0.02
(0.01) (0.02) (0.02) (0.03)
I do not like the foods included in the coupon 0.06 0.03 0.08 -0.04**
(0.01) (0.01) (0.02) (0.02)
Prices are very high at the supermarket 0.66 0.78 0.59 0.19***
(0.02) (0.03) (0.03) (0.04)
Other problems 0.10 0.09 0.10 -0.01
(0.01) (0.02) (0.02) (0.03)
Number of observations 435 156 279
Notes: Mean values reported with standard errors in parentheses below means. * indicates significance at the 10
percent level, ** significance at the 5 percent level, and *** significance at the 1 percent level.
Table 5.7 shows difficulties reported by the food transfer beneficiaries. One out of four
food beneficiaries report problems with packing, including tearing or leakage, and this
percentage is significantly higher in Sucumbíos as compared to Carchi (31 percent versus 19
percent, respectively). A relatively small percentage of beneficiaries report other problems,
including food spoilage (6 percent) or infestation (1 percent), although these issues are again
more likely to be experienced in Sucumbíos, which is not surprising, given the higher
temperature and humidity of food storage facilities in Sucumbíos as compared to Carchi.
Approximately 6 percent of beneficiaries report disliking the food basket composition, while 7
percent cite the bulkiness and weight of the food as being a problem.
Table 5.7 Difficulties experienced by food beneficiaries, by province
Indicator (=1) if true All Carchi Sucumbíos Difference
Torn packages torn / food spills 0.25 0.19 0.31 -0.12**
(0.02) (0.03) (0.04) (0.05)
Food in poor condition / rotten / moldy 0.06 0.03 0.09 -0.06**
(0.01) (0.01) (0.02) (0.03)
Foods infested with insects 0.01 0.00 0.02 -0.02*
(0.01) (0.00) (0.01) (0.01)
Did like foods in the basket (taste, quality, odor, variety, etc.) 0.06 0.06 0.06 0.00
(0.01) (0.02) (0.02) (0.03)
Very large and heavy packages 0.07 0.10 0.05 0.04
(0.01) (0.02) (0.02) (0.03)
Other problems 0.01 0.02 0.01 0.01
(0.01) (0.01) (0.01) (0.01)
Number of observations 334 168 166
Note: Mean values reported with standard errors in parentheses below means. * indicates significance at the 10
percent level, ** significance at the 5 percent level, and *** significance at the 1 percent level.
Finally, Table 5.8 reports problems experienced by the cash transfer households. Overall,
the main difficulties facing beneficiaries are lack of understanding on how to use the debit cards
25
(25 percent) or forgetting passwords (10 percent); however, a number of individuals also report
ATM machine malfunctions or withdrawal of insufficient/wrong amounts (8 percent and 6
percent, respectively). Anecdotally, bank officials stated that a handful of beneficiaries received
less than the $40 monthly transfer because they were charged a minimal fee if they checked
their bank account balance. Some of these difficulties can be expected when introducing
programs using a new financial institution and operating in a population with low financial
literacy.
Table 5.8 Difficulties experienced by cash beneficiaries, by province
Indicator (=1) if true All Carchi Sucumbíos Difference
Insufficient/wrong amount 0.06 0.06 0.07 0.00
(0.01) (0.02) (0.02) (0.02)
Debit card malfunction 0.08 0.06 0.09 -0.03
(0.01) (0.02) (0.02) (0.03)
Malfunction of the ATM machine 0.08 0.07 0.08 -0.01
(0.01) (0.02) (0.02) (0.03)
Did not understand card use 0.25 0.26 0.24 0.02
(0.02) (0.03) (0.03) (0.04)
Forgot password 0.10 0.06 0.12 -0.06**
(0.01) (0.02) (0.02) (0.03)
Other problems 0.03 0.04 0.02 0.02
(0.01) (0.01) (0.01) (0.02)
Number of observations 420 175 245
Notes: Mean values reported with standard errors in parentheses below means. * indicates significance at the 10
percent level, ** significance at the 5 percent level, and *** significance at the 1 percent level.
Tables 5.9 and 5.10 show results for who in the household is reported to have control
over how the transfer is used or spent. Here we split the sample by the gender of the household
head and we focus on differences in decisions made by household head, spouse, joint decisions
between head and spouse, or other individual in the household. Results indicate that the cash
and voucher arms in male-headed households have very similar distributions, with spouses
having most control over spending or use decisions over the transfer (48-50 percent), followed
by the household head (22–25 percent), and then joint decision-making (21-22 percent).
However, for food beneficiary households, 60 percent of spouses make the decisions, compared
to 22 percent of heads and 13 make decisions jointly. This shows that there are some perceived
differences within the household of who should “control” different types of transfers, and
specifically that food has a greater likelihood of being controlled by female spouses. There are
less female headed households in the sample and in these households only 2 percent are
married. Consequently, the main decisions are made either by the head (85-91 percent) or by
others (9-11 percent).
26
Table 5.9 Decision on how to spend or use the transfer in male headed households, by modality
Who decides what to do with the transfer? Food Cash Voucher
Food -
cash
Food -
voucher
Cash -
voucher
Household head 0.22 0.22 0.25 -0.01 -0.03 -0.02
(0.03) (0.02) (0.02) (0.04) (0.04) (0.03)
Household spouse 0.60 0.50 0.48 0.10** 0.12*** 0.02
(0.03) (0.03) (0.03) (0.04) (0.04) (0.04)
Household head and spouse together 0.13 0.22 0.21 -0.09*** -0.08*** 0.01
(0.02) (0.02) (0.02) (0.03) (0.03) (0.03)
Other relative/nonrelative with/without the
head/spouse
0.06 0.06 0.06 0.00 0.00 -0.01
(0.01) (0.01) (0.01) (0.02) (0.02) (0.02)
Number of observations 260 303 308
Notes: Mean values reported with standard errors in parentheses below means. * indicates significance at the 10
percent level, ** significance at the 5 percent level, and *** significance at the 1 percent level.
Table 5.10 Decision on how to spend or use the transfer in female-headed households, by modality
Who decides what to do with the transfer? Food Cash Voucher
Food -
Cash
Food -
Voucher
Cash -
Voucher
Household head 0.91 0.91 0.85 0.00 0.06 0.06
(0.03) (0.03) (0.03) (0.04) (0.04) (0.04)
Household spouse 0.00 0.00 0.01 0.00 -0.01 -0.01
(0.00) (0.00) (0.01) (0.00) (0.01) (0.01)
Household head and spouse together 0.00 0.00 0.03 0.00 -0.03** -0.03**
(0.00) (0.00) (0.01) (0.00) (0.01) (0.01)
Other relative/nonrelative with/without the
head/spouse
0.09 0.09 0.11 0.00 -0.03 -0.02
(0.03) (0.03) (0.03) (0.04) (0.04) (0.04)
Number of observations 81 122 133
Notes: Mean values reported with standard errors in parentheses below means. * indicates significance at the 10
percent level, ** significance at the 5 percent level, and *** significance at the 1 percent level.
In Table 5.11 we show self-reported use of the most recent transfer for food beneficiary
households. As previously noted, this table is descriptive and discrepancies may exist between
self-reports and actual behavior. A more detailed analysis on actual use can be found in Chapter
6, however, this table along with tables 5.12 and 5.13 provide context for why we see differences
in outcomes across modalities. On average, 63 percent of last food transfer was consumed by
the household, and this percentage is higher in Sucumbíos (68 percent) as compared to Carchi
(59 percent). Approximately 29 percent of the food is reported to be saved for consumption in
“harder times” and as might be expected, Carchi beneficiaries report saving a larger percentage
of food as compared to Sucumbíos (34 percent versus 25 percent). Food recipients also report
sharing the transfers with family and friends outside their household (approximately 7 percent
of the last transfer), and this percentage is the highest across all modalities. In general, the share
of the last transfer that beneficiaries report to have sold to buy staples, non-staples, or nonfood
items is very small, of monetary equivalence of approximately $0.27.
27
Table 5.11 Reported uses of last food transfer, by province
Percentage used with: All Carchi Sucumbíos Difference
Own consumption 63.17 58.53 67.86 -9.33***
(1.58) (2.10) (2.31) (3.12)
Sold to buy or trade for staple foods 0.28 0.06 0.51 -0.45
(0.21) (0.06) (0.43) (0.43)
Sold to buy or trade non-staple foods (soda, candy) 0.06 0.06 0.06 0.00
(0.04) (0.06) (0.06) (0.08)
Sold to buy or trade for goods or other nonfood items 0.33 0.12 0.54 -0.42
(0.17) (0.08) (0.33) (0.34)
Shared with family or friends outside the home 6.78 7.10 6.45 0.64
(0.76) (0.97) (1.17) (1.52)
Saved for use in tough times 29.38 34.14 24.57 9.56***
(1.57) (2.17) (2.22) (3.11)
Number of observations 334 168 166
Notes: Mean values reported with standard errors in parentheses below means. * indicates significance at the 10
percent level, ** significance at the 5 percent level, and *** significance at the 1 percent level.
Table 5.12 shows results for similar outcomes among the cash transfer household.
Beneficiaries report expenditures of $32.67 out of the $40 (or 82 percent) on staple foods
including rice, beans, and other items, $3.33 (or 8 percent) on savings, and $2.50 (or 6 percent)
on nonfood expenses. Very little of the transfer was reported to be spent on non-staple foods
($0.54) or shared with family or friends outside the household ($0.95). Beneficiaries in Carchi
report spending significantly more (on average, $2.41 more) on staple foods; however, this may
reflect higher food prices rather than higher relative expenditures.
Finally, voucher beneficiaries also use the majority of their coupon to purchase staple
foods ($23.65 or 60 percent). Meat and eggs make up the next highest expenditure category
($8.84 or 22 percent) followed by fruits and vegetables ($7.04 or 17 percent). Although the
amount spent on staple foods is lower than what is reported for the cash households, voucher
beneficiaries spend almost all their money ($39.5) on staple foods, fruits and vegetables, and
meats and eggs (Table 5.13). There is virtually no reported selling, trading, or sharing of
vouchers.
Table 5.12 Reported uses of last cash transfer, by province
Amount spent on ($): All Carchi Sucumbíos Difference
Staple foods (rice, beans, etc.). 32.67 34.08 31.67 2.41***
(0.51) (0.75) (0.68) (1.02)
Non-staple foods (soda, candy) 0.54 0.73 0.40 0.33
(0.14) (0.21) (0.18) (0.28)
Nonfood or other expenses 2.50 1.79 3.02 -1.23**
(0.30) (0.46) (0.40) (0.61)
Share with family or friends outside the home 0.95 1.00 0.92 0.08
(0.23) (0.39) (0.27) (0.48)
Save for later use 3.33 2.40 4.00 -1.60**
(0.35) (0.45) (0.50) (0.67)
Number of observations 420 175 245
Notes: Mean values reported with standard errors in parentheses below means. * indicates significance at the 10
percent level, ** significance at the 5 percent level, and *** significance at the 1 percent level.
28
Table 5.13 Reported use of vouchers, by province
Amount used with: All Carchi Sucumbíos Difference
Staple foods (rice, beans, etc.). 23.65 22.27 24.42 -2.15***
(0.36) (0.51) (0.47) (0.70)
Fruits and vegetables 7.04 8.38 6.29 2.09***
(0.23) (0.37) (0.28) (0.47)
Meat, eggs 8.84 8.67 8.93 -0.26
(0.24) (0.35) (0.33) (0.48)
Sold to buy or trade non-staple foods (soda, candy) 0.03 0.03 0.04 0.00
(0.03) (0.03) (0.04) (0.05)
Sold to buy or trade for goods or other nonfood items 0.01 0.03 0.00 0.03
(0.01) (0.03) (0.00) (0.03)
Share with family or friends outside the home 0.43 0.61 0.32 0.29
(0.11) (0.26) (0.10) (0.28)
Number of observations 435 156 279
Notes: Mean values reported with standard errors in parentheses below means. * indicates significance at the 10
percent level, ** significance at the 5 percent level, and *** significance at the 1 percent level.
5.2 Nutrition Trainings
In this section we provide results surrounding nutrition trainings and nutrition
knowledge of beneficiaries and non-beneficiaries. We start by describing attendance and
experience with monthly nutritional trainings in Table 5.14. Because transfers were conditioned
on attendance, it is not surprising that attendance rates are very high: on average, beneficiaries
report attending 5.65 of 6 trainings (average of 5.8 in Carchi and 5.55 in Sucumbíos).
Beneficiaries are also using and sharing the information learned at the trainings: 96 percent
report directly using and 89 percent reported sharing information learned with friends and/or
neighbors.
Table 5.14 Nutrition training sessions
Overall: All Carchi Sucumbíos Difference
Number of training sessions attended 5.65 5.80 5.55 0.26***
(0.03) (0.03) (0.04) (0.05)
Share training information with friends and/or neighbors? 0.89 0.88 0.90 -0.02
(0.01) (0.01) (0.01) (0.02)
Has ever put into practice what is taught in the training 0.96 0.95 0.97 -0.02***
(0.01) (0.01) (0.01) (0.01)
Number of observations 1,194 499 695
Notes: Mean values reported with standard errors in parentheses below means. * indicates significance at the 10
percent level, ** significance at the 5 percent level, and *** significance at the 1 percent level. Sample is composed of
beneficiaries that attended at least one session. Of the 1207 beneficiaries with nonmissing data, only 13 (or 1 percent)
never attended a training.
5.3 Nutrition Knowledge
As mentioned in Chapter 2 of the report, the nutrition sessions included information on
program sensitization, family nutrition, food and nutrition for pregnant and lactating women,
29
nutrition for children aged 0–12 months, and nutrition for children aged 12–24 months. In order
to investigate whether the nutrition sessions increased participant’s knowledge, we analyze
questions on nutrition knowledge collected at baseline and follow-up. Because we want to
observe changes in knowledge among the same individuals, we restrict the sample to 1,880 of
the 2,122 households where the same individuals were administered the survey between
rounds.
Table 5.15 reveals that 77 percent of respondents at baseline know that breastfeeding
should start immediately after the child is born; 60 percent know that a baby should start eating
food at 6 months; and 71 percent know that a one-year-old should not only be eating the same
things as the rest of the family. For these indicators, there are no significant differences between
treatment and comparison households. The surveys also asked respondents to name the
advantages of breastfeeding; food and beverages pregnant woman should abstain from eating
and drinking; food items rich in vitamin A and iron; and ways to treat water. For each question,
there is a total number of correct items that could have been answered. For example, for
breastfeeding, participants could name up to three advantages of breastfeeding. Table 5.15
reveals that on average, respondents are able to name 1.6 advantages of breastfeeding, 1.9
food/beverages pregnant woman should not eat/drink, 0.9 iron-rich food items, 0.6 vitamin A-
rich food items, and 1.2 ways to treat drinking water. For correct answers to iron-rich food
items, the treatment arm has a significantly higher average and for correct answers to items
pregnant woman should not eat/drink, the comparison arm has a significantly higher average at
baseline.
Table 5.15 Baseline means, by pooled treatment
All Comparison Treatment Difference
How soon should baby start breast feeding: Immediately 0.77 0.77 0.77 -0.00
(0.01) (0.02) (0.01) ( 0.02)
At what age should baby start eating food besides breast
milk: 6 months
0.60 0.60 0.61 0.00
(0.01) (0.02) (0.01) ( 0.03)
Should 1-year-old child only eat same thing as rest of family?
No
0.71 0.71 0.72 0.01
(0.01) (0.02) (0.01) ( 0.02)
Mean number of items named for:
Advantages of breast feeding (0-3) 1.64 1.67 1.63 -0.04
(0.02) (0.04) (0.02) ( 0.04)
Items pregnant mothers should not eat or drink (0-4) 1.91 1.99 1.88 -0.10
(0.02) (0.05) (0.03) ( 0.06)*
Iron-rich food items (0-7) 0.94 0.88 0.97 0.09
(0.02) (0.05) (0.03) ( 0.05)*
Vitamin A-rich food items (0-5) 0.63 0.60 0.64 0.04
(0.02) (0.03) (0.02) ( 0.04)
Ways to treat water for drinking (0-5) 1.19 1.18 1.20 0.02
(0.01) (0.02) (0.01) ( 0.03)
N 1,880 507 1,373
Notes: Standard errors in parentheses. * p < 0.1 ** p < 0.05; *** p < 0.01.
In order to see if there are improvements in participants’ nutrition knowledge, we ask
the same questions at follow-up. Table 5.16 shows that the percentage of respondents that know
30
that an infant should begin breastfeeding immediately, start eating food at 6 months, and that a
1-year-old child should not only be eating the same food as the rest of the family increases from
baseline to follow-up. Although we find slight declines in the correct answers for the number of
items named for advantages of breast-feeding and food/drinks pregnant woman should not eat
or drink, we find large increases from baseline to follow-up on the number of items named for
iron-rich food sources; vitamin A-rich food sources, and ways to treat water. At follow-up there
are also large and significant differences in means across treatment and comparison arms. In
particular, a larger percentage of those in the treatment arm know that an infant should start
eating food (other than breast milk) at 6 months, and those in the treatment arm can name
significantly more correct iron- and vitamin A-rich food sources. Given that baseline knowledge
on feeding practices and prenatal care are already relatively high, it is not surprising that the
increases in knowledge are mainly concentrated on nutritious food items. In summary,
nutrition knowledge increased from baseline to follow-up for both treatment and comparison
groups for 6 out of the 8 questions listed. The small increases in the comparison mean suggest
spillover effects and knowledge sharing. In general, however, the increases were much larger
for the treatment group.
Table 5.16 Follow-up means, by pooled treatment
All Comparison Treatment Difference
How soon should baby start breast feeding: Immediately 0.80 0.79 0.81 0.02
(0.01) (0.02) (0.01) ( 0.02)
At what age should baby start eating food besides breast
milk: 6 months
0.71 0.66 0.73 0.07
(0.01) (0.02) (0.01) ( 0.02)***
Should 1-year-old child only eat same thing as rest of
family? No
0.75 0.77 0.74 -0.03
(0.01) (0.02) (0.01) ( 0.02)
Mean number of items named for:
Advantages of breast feeding (0-3) 1.56 1.55 1.57 0.01
(0.02) (0.03) (0.02) ( 0.04)
Items pregnant mothers should not eat or drink (0-4) 1.89 1.85 1.90 0.05
(0.02) (0.04) (0.02) ( 0.05)
Iron-rich food items (0-7) 1.42 1.04 1.56 0.52
(0.02) (0.04) (0.03) ( 0.05)***
Vitamin A-rich food items (0-5) 0.85 0.63 0.93 0.30
(0.02) (0.03) (0.02) ( 0.04)***
Ways to treat water for drinking (0-5) 1.25 1.24 1.25 0.01
(0.01) (0.02) (0.01) ( 0.03)
N 1,880 507 1,373
Notes: Standard errors in parentheses. * p < 0.1 ** p < 0.05; *** p < 0.01.
31
6. Impact on Food Consumption and Dietary Diversity
6.1 Indicators and Descriptive Statistics
Household value of food consumption aggregates are constructed from data on the total
value of food consumed in the last seven days. These data are asked with reference to 41
different food items. Aggregates are constructed using not just food purchased in the market
place but also food that is home-produced, food that is received as gifts or remittances from
other households or institutions, and food that is received as payments for in-kind services.
Median prices from food purchased are inputted to calculate the total value of food consumed
from home production or received as gift or in-kind payment. Weekly household values on
food consumed are converted to monthly values which are then converted to household per
capita values by dividing by the number of members in the household. Given that the
distribution of per capita food consumption is skewed to the right, we convert all values to their
logarithms for the analysis, and we drop outliers by trimming the top and bottom .5 percent of
the distribution. We also convert to missing observations that report no food consumption in
the last week.
Caloric intake is constructed from the amount of food consumed by households (from
purchases, own stock, or in kind payments). In particular, the amount of food consumed for
each item is multiplied by the energy value for that item to obtain the kilocalories consumed.
Energy values are taken from the Nutrition Database for Standard Reference (USDA 2010) and
from the Tabla de Composicion de Alimentos de Centroamerica (Manchu and Mendez 2007). Total
monthly household caloric values are then converted to daily amounts and divided by
household size to obtain caloric availability per person per day. Similar to consumption
aggregates, all values are converted to their logarithms, and outliers at the top and bottom .5
percent of the distribution are converted to missing as are values that report no food
consumption in the last week.
Food consumption and caloric intake play important roles in meeting food security
needs. However, households do not solely value quantity – a more varied diet is also important.
Increased dietary diversity is associated with a number of improved outcomes in areas such as
birth weight, child anthropometrics, hemoglobin concentrations, hypertension, cardiovascular
disease, and cancer (Hoddinott and Yohannes 2002). We construct three separate measures for
dietary quality: the Dietary Diversity Index (DDI), Household Dietary Diversity Score (HDDS),
and the Food Consumption Score (FCS). The most straightforward of these measures, the
Dietary Diversity Index, sums the number of distinct food items consumed by the household in
the previous seven days. The household questionnaire covers 41 such food items, and thus the
DDI in this survey can feasibly range from 0 (no consumption at all) to 41. Hoddinott and
Yohannes (2002) show that the DDI correlates well with both household dietary quantity and
quality, and thus provides a useful summary point of comparison within the measured sample.
The HDDS captures a similar element of food access, although it differs from DDI in that
frequency is measured across standardized food groups, instead of individual food items. The
score is calculated by summing the number of food groups consumed in the previous seven
days from the following 12 groups (Kennedy, Ballard, and Dop 2011): cereals, roots/tubers,
vegetables, fruits, meat/poultry/offal, eggs, fish/seafood, pulses/legumes/nuts, milk/milk
32
products, oils/fats, sugar/honey, miscellaneous. Lastly, WFP measures food insecurity using a
proxy indicator called the food consumption score (FCS). The FCS is calculated by summing the
number of days eight different food groups (staples, pulses, vegetables, fruit, meat/fish,
milk/dairies, sugar/honey, oils/fats) were consumed by a household, multiplying these by
weighted frequencies and summing across categories to obtain a single proxy indicator (Table
6.1). The FCS has been found to correlate well with caloric availability at the household level
(Wiesmann, Bassett et al. 2009) and thus reflects the quality of the diet in terms of energy and
diversity. Across all three dietary diversity measures, DDI, is the most highly correlated with
the value of food consumption, caloric intake, total household expenditures, and the wealth
index. Given that it is very unlikely that a household consumed zero food items in the last
week, for our analysis we convert observations from these households to missing (this occurred
for approximately 1 percent of the sample).
Table 6.1 Aggregate food groups and weights to calculate the Food Consumption Score
Group Food items Food group Weight
1 Maize, maize porridge, rice, sorghum, millet past, bread and other cereals
Staples 2 Cassava, potatoes and sweet potatoes, other tubers, plantains
2 Beans, peas, groundnuts and cashew nuts Pulses 3
3 Vegetables, leaves Vegetables 1
4 Fruits Fruit 1
5 Beef, goat, poultry, pork, eggs, and fish Meat and fish 4
6 Milk, yogurt, and other dairies Milk 4
7 Sugar, sugar products, and honey Sugar 0.5
8 Oils, fats, and butter Oil 0.5
Source: WFP (2008).
Tables 6.2 and 6.3 show the mean values at baseline and follow-up of the food
consumption and dietary diversity measures - value of per capita food consumption, per capita
caloric intake, HDDS, DDI, and FCS. At baseline, households consume approximately $39 of
food per capita, 1875 kilocalories per capita, 9 out of the 12 food groups (HDDS), 17 out of the
40 food items (DDI)2, and have a food frequency score (FCS) of 60. For the value of per capita
food consumption and per capita caloric intake, the treatment arm has significantly higher
means. However, there are no significant differences in means across treatment and comparison
arms for the dietary diversity measures. At follow-up, per capita food consumption and caloric
intake only increase for the treatment arm and not the comparison arm, thus the difference in
means across treatment and comparison arm increases. For dietary diversity measures, the
average HDDS score increases to nearly 11 food groups, the DDI increases to 21 food items, and
the FCS to nearly 68. Both treatment and comparison arms experience increases in their HDDS
DDI, and FCS scores; however, the increase for the treatment arm is much larger. For all three
indices, the treatment arm has significantly higher means than the comparison arm at follow-
up, suggesting large impacts of the intervention.
2 At baseline data on consumption of oils and fats was not collected, thus there are only 40 food items at
baseline and 41 at follow-up.
33
Table 6.2 Baseline food consumption and dietary diversity measures, by treatment status
Outcome variables All Comparison Treatment Difference
Per capita food consumption (monthly) 39.28 37.03 40.13 3.10
(0.62) (1.12) (0.74) ( 1.34)**
N 1,985 545 1,440
Caloric intake per capita (daily) 1,874.93 1,794.06 1,905.40 111.34
(25.20) (45.40) (30.15) (54.50)**
N 2,006 549 1,457
In last 7 days:
Number of food groups (0-12) consumed – HDDS 9.17 9.10 9.20 0.09
(0.04) (0.07) (0.04) ( 0.08)
Number of unique foods (0-40) consumed – DDI 17.31 17.04 17.41 0.37
(0.12) (0.24) (0.14) ( 0.28)
Weighted food group frequency (0-112) consumed – FCS 60.24 59.45 60.54 1.09
(0.44) (0.87) (0.51) ( 1.00)
N 2,087 562 1,525
Notes: Mean values reported with standard errors in parentheses below means. * indicates significance at the 10
percent level, ** significance at the 5 percent level, and *** significance at the 1 percent level.
Table 6.3 Follow-up food consumption and dietary diversity measures, by treatment status
Outcome variables All Comparison Treatment Difference
Per capita food consumption (monthly) 42.67 36.70 44.93 8.22
(0.57) (0.95) (0.69) ( 1.18)***
N 1,985 545 1,440
Caloric intake per capita (daily) 1,953.22 1,748.86 2,030.22 281.36
(23.44) (40.96) (28.08) (49.66)***
N 2,006 549 1,457
In last 7 days: Number of food groups (0-12) consumed – HDDS 10.76 10.35 10.91 0.56
(0.03) (0.07) (0.04) ( 0.08)***
Number of unique foods (0-40) consumed – DDI 21.24 19.13 22.02 2.89
(0.13) (0.24) (0.15) ( 0.28)***
Weighted food group frequency (0-112) consumed – FCS 67.90 62.13 70.02 7.89
(0.44) (0.79) (0.52) ( 0.94)***
N 2,087 562 1,525
Notes: Mean values reported with standard errors in parentheses below means. * indicates significance at the 10
percent level, ** significance at the 5 percent level, and *** significance at the 1 percent level.
Figure 6.1 shows the distributions the indices and reveals a similar pattern across
treatment and comparison households from baseline to follow-up. In particular, the densities of
the treatment and comparison arms for the log value of per capita food consumption (top panel),
DDI (middle panel) and FCS (bottom panel) are very similar at baseline, but at follow-up the
densities for the treatment arm have shifted to the right of the comparison arm, indicating higher
values at all ends of the distribution.
34
Figure 6.1 Density graphs of food consumption and dietary diversity indices, by treatment and
control arms at baseline and follow-up
6.2 Impacts of Pooled Treatment on Food Consumption and Dietary Diversity
In Tables 6.4 and 6.5 we combine all three treatment arms (food, cash, and voucher) and
estimate the impact of pooled treatment on the value of per capita food consumption, per capita
caloric intake, HDDS, DDI, and FCS. We present the ANCOVA estimates with and without
controlling for covariates.
0.2
.4.6
.8
1 2 3 4 5 6Log value of per capita food consumption
Comparison Treatment
Baseline
Log per capita food consumption
0.2
.4.6
.8
1 2 3 4 5 6Log value of per capita food consumption
Comparison Treatment
Follow-up
Log per capita food consumption
0
.02
.04
.06
.08
0 10 20 30 40DDI
Comparison Treatment
Baseline
Dietary Diversity Index
0
.02
.04
.06
.08
0 10 20 30 40DDI
Comparison Treatment
Follow-up
Dietary Diversity Index
0
.00
5.0
1.0
15
.02
0 35 50 100FCS
Comparison Treatment
Baseline
Food Consumption Score
0
.00
5.0
1.0
15
.02
0 35 50 100FCS
Comparison Treatment
Follow-up
Food Consumption Score
35
For food consumption and caloric intake we convert the values to their logarithms, and
thus, the coefficients in Table 6.4 can be interpreted as percent changes. Focusing on results
from columns 2 and 4, we find that being in the pooled treatment arm leads to a 13 percent
increase in a household’s value of per capita food consumption and a 10 percent increase in per
capita caloric intake. While the results are significant, the magnitude is slightly lower than what
we would expect to see if households used the whole transfer on food. In particular, given that
the average household size is approximately four, we would expect to see a per capita increase
of $10, which is approximately a 24 percent increase from baseline per capita food consumption
values. Thus, our findings suggest that not all the transfer is being used for immediate food
consumption which is consistent with the self-reported data presented earlier.
Table 6.4 Impact of pooled treatment on food consumption measures, with and without covariates
Log value of per capita
food consumption
Log per capita caloric
intake
(1) (2) (3) (4)
Pooled Treatment 0.17 0.13 0.13 0.10
(0.03)*** (0.03)*** (0.03)*** (0.03)***
Household head is Colombian -0.06 -0.04
(0.03)** (0.03)
Household head is female -0.01 -0.01
(0.02) (0.02)
Age of household head 0.00 0.00
(0.00)*** (0.00)***
Household head has at least secondary education 0.04 -0.01
(0.02)* (0.02)
Number of children 0-5 years 0.01 -0.02
(0.02) (0.02)
Number of children 6-15 years 0.01 0.00
(0.01) (0.01)
Household size -0.11 -0.10
(0.01)*** (0.01)***
Wealth index: 2nd quintile 0.03 0.03
(0.03) (0.03)
Wealth index: 3rd quintile 0.03 0.01
(0.03) (0.03)
Wealth index: 4th quintile 0.04 0.04
(0.04) (0.04)
Wealth index: 5th quintile 0.06 0.02
(0.04)* (0.03)
Baseline log value of per capita food consumption 0.43 0.29
(0.02)*** (0.02)***
Baseline log per capita caloric intake 0.38 0.25
(0.02)*** (0.02)***
Constant 1.98 2.77 4.51 5.82
(0.08)*** (0.10)*** (0.16)*** (0.19)***
R2 0.25 0.35 0.20 0.31
N 1,985 1,985 2,006 2,006
Notes: Standard errors in parentheses clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. Columns 2 and 4
contain urban center fixed effects.
36
In addition to significantly increasing the quantity of food consumed, the program
increases the quality of food consumed as measured by the dietary diversity outcomes (Table
6.5). In particular, HDDS increases by 0.47 points, which is a 5.1 percent increase from the
baseline mean; DDI increases by 2.49 points, which is a 14.4 percent increase from the baseline
mean; and FCS increases by 7.57 points, which is a 12.6 percent increase from the baseline mean
(columns 2, 4, and 6). Other factors that contribute significantly to the dietary diversity
measures are whether the household head is Colombian and wealth.
Table 6.5 Impact of pooled treatment on dietary diversity measures, with and without covariates
HDDS DDI FCS
(1) (2) (3) (4) (5) (6)
Pooled Treatment 0.53 0.47 2.71 2.49 7.51 7.56
(0.11)*** (0.10)*** (0.37)*** (0.41)*** (1.06)*** (1.14)***
Household head is Colombian -0.27 -1.03 -1.91
(0.08)*** (0.27)*** (0.95)**
Household head is female 0.08 0.29 0.40
(0.07) (0.27) (0.84)
Age of household head -0.01 -0.00 -0.02
(0.00)* (0.01) (0.03)
Household head has at least secondary education -0.04 0.37 2.17
(0.07) (0.27) (0.89)**
Number of children 0-5 years 0.00 0.08 0.01
(0.05) (0.18) (0.78)
Number of children 6-15 years -0.07 -0.26 -0.54
(0.04)* (0.16) (0.58)
Household size 0.07 0.22 0.68
(0.03)** (0.12)* (0.41)*
Wealth index: 2nd quintile 0.13 0.24 2.11
(0.11) (0.35) (1.49)
Wealth index: 3rd quintile 0.21 0.52 3.14
(0.12)* (0.36) (1.43)**
Wealth index: 4th quintile 0.22 0.86 3.03
(0.12)* (0.40)** (1.73)*
Wealth index: 5th quintile 0.31 0.80 4.23
(0.11)*** (0.35)** (1.49)***
Baseline household dietary diversity score 0.33 0.28
(0.03)*** (0.03)***
Baseline dietary diversity index 0.48 0.43
(0.02)*** (0.02)***
Baseline food consumption score 0.35 0.31
(0.03)*** (0.02)***
Constant 7.36 7.73 10.91 11.43 41.17 40.23
(0.30)*** (0.32)*** (0.51)*** (0.73)*** (1.72)*** (2.88)***
R2 0.16 0.19 0.26 0.29 0.15 0.18
N 2,087 2,087 2,087 2,087 2,087 2,087
Notes: Standard errors in parentheses clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. Columns 2, 4, and
6 contain urban center fixed effects.
37
6.3 Impact by Treatment Arm on Food Consumption and Dietary Diversity
Tables 6.6 and 6.7 present the ANCOVA estimate for each treatment arm separately, and
conduct Wald tests to examine whether the estimates from each treatment arm are significantly
different from each other. Specifications include a full set of control variables; however, for
simplicity we only present the coefficients from the different treatment arms. Table 6.6 shows
that all three treatment arms lead to significant increases in the value of per capita food
consumption that range from 12-16 percent. As revealed from the F-tests at the bottom of the
table, there are no statistically significant differences across treatment arms in the size of the
impact. Similarly, all three treatment arms lead to significant increases in caloric intake that
range from 6-16 percent. However, the impact of food on per capita caloric intake is
significantly larger than that of the cash transfer.
Table 6.6 Impact of treatment modalities on food consumption measures
Log value of per capita
food consumption
Log per capita caloric
intake
Treatment==Food 0.16 0.16
(0.04)*** (0.04)***
Treatment==Cash 0.12 0.06
(0.04)*** (0.03)*
Treatment==Voucher 0.13 0.11
(0.04)*** (0.03)***
R2 0.35 0.32
N 1,985 2,006
F test: Food=Voucher 0.79 2.67
P-value 0.38 0.10
F test: Cash=Voucher 0.16 1.82
P-value 0.69 0.18
F test: Food=Cash 1.49 6.99
P-value 0.22 0.01
Notes: Standard errors in parenthesis clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01.
All estimations contain baseline outcome variable, urban center fixed effects, and a full set of
control variables.
All three modalities (food, cash, and voucher) significantly increase the three measures
of dietary diversity; however, the size of the increase differs by treatment arm (Table 6.7). In
particular, the voucher leads to significantly larger impacts than food for the DDI and
significantly larger impacts than food and cash for the FCS measure. In order to better
understand the magnitude of the impact of treatment on dietary diversity outcomes, we convert
unit changes from Table 6.7 to percent changes from baseline means and present these changes
graphically. Figure 6.2 reveals that the percentage increase in HDDS is small compared to the
percentage increase in DDI and FCS. In particular, HDDS increases by 5.6 percent for the food
and voucher group, and by 4.4 percent for the cash group. Vouchers lead to the largest
percentage increase in DDI and FCS measures: 16.7 percent increase in DDI compared to 11.4
percent and 13.8 percent increase for the food and cash group respectively, and a 15.6 percent
increase in FCS compared to a 10.1 percent and 10.8 percent increase for food and cash
households.
38
Table 6.7 Impact of treatment modalities on dietary diversity measures
HDDS DDI FCS
Treatment==Food 0.51 1.98 6.10
(0.12)*** (0.50)*** (1.46)***
Treatment==Cash 0.40 2.39 6.48
(0.11)*** (0.44)*** (1.34)***
Treatment==Voucher 0.51 2.89 9.41
(0.11)*** (0.46)*** (1.36)***
R2 0.19 0.29 0.19
N 2,087 2,087 2,087
F test: Food=Voucher 0.00 4.34 5.08
P-value 0.99 0.04 0.03
F test: Cash=Voucher 1.85 1.83 4.66
P-value 0.18 0.18 0.03
F test: Food=Cash 1.47 1.02 0.08
P-value 0.23 0.31 0.78
Notes: Standard errors in parentheses clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01.
All estimations contain baseline outcome variable, urban center fixed effects, and a full set of
control variables.
Figure 6.2 Percent increase in dietary diversity measures, by treatment arms
* p < 0.1 ** p < 0.05; *** p < 0.01.
Tables 6.5 and 6.7 reveal that the food, cash, and voucher program leads to significant
increases in the variety of foods consumed. In order to provide a better understanding of how
this impacts households on the bottom end of the distribution in terms of the FCS, we analyze
whether the program leads to a decrease in the percentage of households classified as having
“poor to borderline food consumption.” According to the WFP guidelines, households are
categorized as having poor to borderline health if their FCS score is less than or equal to 35
5.6***
11.4***
10.1***
4.4***
13.8***
10.8***
5.6***
16.7*** 15.6***
0.0
2.0
4.0
6.0
8.0
10.0
12.0
14.0
16.0
18.0
HDDS DDI FCS
%
Food
Cash
Voucher
39
(Kennedy, Ballard, and Dop 2011). At baseline, approximately 10.5 percent of households have
poor to borderline health. This is comparable to the 8.3 percent found in the WFP’s assessment
of the food security and nutrition situation among Colombian refugees (PMA 2010). Table 6.8
presents the estimates of the impact of treatment on an indicator for having poor to borderline
food consumption, and reveals that overall pooled treatment significantly decreases poor to
borderline food consumption by 4 percentage points, reducing the proportion of households
with poor or borderline food consumption by approximately 40 percent. The second column of
Table 6.8 shows that the food and voucher arm significantly decrease the percent of households
with poor to borderline consumption by 6 and 4 percentage points respectively. The impact of
the food arm is significantly different to that of the cash arm.
Table 6.8 Impact of treatment on poor to borderline food consumption
Outcome variable: =1 if poor to
borderline food consumption
Pooled Treatment -0.04
(0.02)***
Treatment==Food -0.06
(0.02)***
Treatment==Cash -0.03
(0.02)
Treatment==Voucher -0.04
(0.02)**
R2 0.08 0.08
N 2,087 2,087
F test: Food=Voucher 1.74
P-value 0.19
F test: Cash=Voucher 1.53
P-value 0.22
F test: Food=Cash 5.97
P-value 0.02
Notes: Standard errors in parentheses clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01.
All estimations contain baseline outcome variable, urban center fixed effects, and a full set of
control variables.
6.4 Impacts by Food Groups
Tables 6.6 and 6.7 show that all three modalities lead to improvements in a household’s
diet, however, food leads to significantly larger increases in caloric intake, and voucher leads to
significantly larger increases in dietary diversity. In order to investigate what food groups are
increasing as a result of being in the treatment arms, we conduct our estimations separately for
the 12 food groups included in the HDDS. In particular, we conduct ANCOVA estimates on
food frequencies—number of days in the last seven days that a household consumed the
specific food group – and per capita caloric intake.
We find that pooled treatment significantly increases the number of days a household
consumes the following nine food groups: cereals; roots and tubers; vegetables; fruits; meat and
poultry; eggs; fish and seafood; pulses, legumes, and nuts; and milk and dairy (Table 6.9).
Looking at treatment arms separately (Table 6.10), we find that the food transfer leads to
significant increases in the following five groups: cereals; roots and tubers; meat and poultry;
40
fish and seafood; and pulses, legumes, and nuts. This is not surprising, given that the food
baskets consisted of rice, lentils, sardines, and vegetable oil. The cash transfer leads to
significant increases in the following seven food groups: roots and tubers; vegetables; meat and
poultry; eggs; fish and seafood; pulses, legumes, and nuts; and milk and dairy. Finally,
consistent with the previous results, the voucher leads to significant increases in the most
number of food groups, which are the following nine groups: cereals, roots and tubers;
vegetables; fruits; meat and poultry; eggs; fish and seafood; pulses, legumes, and nuts; and milk
and dairy. The impact of vouchers on the frequency of consumption is significantly different to
that of food transfers for vegetables, eggs, and milk and dairy.
In Table 6.11 values are again converted to their logarithms, and thus, the coefficients
can be interpreted as percent changes. Table 6.11 reveals why we find such a large increase in
per capita caloric intake for the food group and not the cash group. In particular, the food group
leads to significantly larger increase in caloric intake from cereals. Calories from cereals at
baseline are 763 kcals which accounts for the largest portion of total calories or 41 percent. Thus
an 18 percent increase in these calories from the food group is equivalent to an increase of 137
kcals, while an 8 percent increase from the cash group is equivalent to 61 kcals. Food also leads
to significantly larger increases than cash in caloric intake from fish and seafood and pulses,
legumes, and nuts.
41
Table 6.9 Impact of pooled treatment on food frequency, by food groups
Outcome variable: Number of days in last 7 days household consumed...
Cereals Roots
&
Tubers
Vegetables Fruits Meat &
poultry
Eggs Fish &
seafood
Pulses
legumes
& nuts
Milk &
dairy
Sugar &
honey
Other Oils &
fats
Pooled
Treatment
0.28 0.49 0.29 0.23 0.28 0.29 0.37 0.84 0.64 -0.00 0.18 -0.09 (0.09)*** (0.14)*** (0.09)*** (0.12)** (0.08)*** (0.12)** (0.08)*** (0.10)*** (0.15)*** (0.09) (0.16) (0.07)
R2 0.08 0.11 0.08 0.12 0.15 0.07 0.08 0.10 0.18 0.04 0.08 0.04
N 2,087 2,087 2,087 2,087 2,087 2,087 2,087 2,087 2,087 2,087 2,087 2,087
Notes: Standard errors in parentheses clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. All estimations contain baseline outcome variable, urban center
fixed effects, and a full set of control variables. Oils and fats were not included in the baseline survey and thus we do not control for it in the estimation.
Table 6.10 Impact of treatment arms on food frequency, by food groups
Outcome variable: Number of days in last 7 days household consumed...
Cereals Roots
&
Tubers
Vegetables Fruits Meat &
poultry
Eggs Fish &
seafood
Pulses
legumes
& nuts
Milk &
dairy
Sugar
&
honey
Other Oils
& fats
Treatment==Food 0.43 0.42 0.13 0.26 0.19 0.06 0.61 1.20 0.19 0.05 0.12 0.00
(0.10)*** (0.19)** (0.12) (0.17) (0.09)** (0.16) (0.12)*** (0.15)*** (0.20) (0.10) (0.18) (0.11)
Treatment==Cash 0.15 0.45 0.30 0.15 0.34 0.26 0.15 0.59 0.66 -0.05 0.24 -0.10
(0.10) (0.17)*** (0.11)*** (0.15) (0.11)*** (0.16)* (0.08)* (0.12)*** (0.17)*** (0.11) (0.17) (0.08)
Treatment==Voucher 0.30 0.56 0.39 0.29 0.28 0.46 0.40 0.83 0.90 0.01 0.16 -0.13
(0.10)*** (0.17)*** (0.10)*** (0.14)** (0.10)*** (0.15)*** (0.09)*** (0.11)*** (0.18)*** (0.10) (0.19) (0.08)
R2 0.09 0.12 0.08 0.12 0.15 0.07 0.10 0.11 0.18 0.04 0.08 0.04
N 2,087 2,087 2,087 2,087 2,087 2,087 2,087 2,087 2,087 2,087 2,087 2,087
F test: Food=Voucher 2.59 0.54 5.44 0.03 0.86 5.00 3.26 6.44 13.87 0.30 0.06 1.57
P-value 0.11 0.46 0.02 0.87 0.36 0.03 0.07 0.01 0.00 0.59 0.81 0.21
F test: Cash=Voucher 2.58 0.42 0.78 0.96 0.21 1.46 9.97 3.67 2.12 0.34 0.28 0.11
P-value 0.11 0.52 0.38 0.33 0.64 0.23 0.00 0.06 0.15 0.56 0.59 0.74
F test: Food=Cash 9.76 0.03 2.40 0.48 1.95 1.34 17.88 17.29 7.21 1.27 0.79 0.92
P-value 0.00 0.87 0.12 0.49 0.17 0.25 0.00 0.00 0.01 0.26 0.38 0.34
Notes: Standard errors in parentheses clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. All estimations contain baseline outcome variable, urban center
fixed effects, and a full set of control variables. Oils and fats were not included in the baseline survey and thus we do not control for it in the estimation.
42
Table 6.11 Impact of treatment arms on log per capita caloric intake, by food groups
Outcome variable: Log per capita caloric intake (daily) ...
Cereals Roots
&
Tubers
Vegetables Fruits Meat &
poultry
Eggs Fish &
seafood
Pulses
legumes
& nuts
Milk &
dairy
Sugar
&
honey
Other Oils &
fats
Treatment==Food 0.18 0.29 0.12 0.14 0.26 0.04 1.08 0.89 0.31 -0.01 0.19 0.05
(0.05)*** (0.12)** (0.06)* (0.08) (0.12)** (0.10) (0.18)*** (0.15)*** (0.17)* (0.11) (0.12)* (0.10)
Treatment==Cash 0.08 0.22 0.12 0.09 0.35 -0.00 0.30 0.38 0.50 0.04 0.06 -0.12
(0.05) (0.12)* (0.06)** (0.08) (0.11)*** (0.09) (0.13)** (0.13)*** (0.14)*** (0.09) (0.11) (0.08)
Treatment==Voucher 0.11 0.22 0.13 0.16 0.31 0.11 0.43 0.59 0.70 0.06 -0.05 -0.07
(0.05)** (0.11)* (0.05)** (0.08)** (0.11)*** (0.09) (0.13)*** (0.13)*** (0.14)*** (0.09) (0.11) (0.06)
R2 0.11 0.06 0.20 0.14 0.18 0.05 0.10 0.10 0.17 0.07 0.09 0.04
N 2,006 2,006 2,006 2,006 2,006 2,006 2,006 2,006 2,006 2,006 2,006 2,006
F test: Food=Voucher 2.29 0.39 0.00 0.16 0.16 0.65 19.63 5.18 7.03 0.57 5.03 1.54
P-value 0.13 0.54 0.95 0.69 0.69 0.42 0.00 0.02 0.01 0.45 0.03 0.22
F test: Cash=Voucher 0.35 0.00 0.00 1.37 0.19 1.64 1.92 3.32 2.99 0.06 1.17 0.33
P-value 0.55 1.00 0.99 0.24 0.67 0.20 0.17 0.07 0.09 0.81 0.28 0.56
F test: Food=Cash 4.10 0.41 0.00 0.43 0.65 0.12 27.18 13.79 1.63 0.32 1.52 2.39
P-value 0.04 0.52 0.97 0.52 0.42 0.72 0.00 0.00 0.20 0.57 0.22 0.12
Notes: Standard errors in parentheses clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. All estimations contain baseline outcome variable, urban center
fixed effects, and a full set of control variables. Oils and fats were not included in the baseline survey and thus we do not control for it in the estimation.
43
6.5 Impacts by Nationality
In Tables 6.12 - 6.15 we explore whether the impact of treatment differed by Ecuadorian
and Colombian households. We do this by creating an interaction term of the treatment
indicator with the indicator for whether or not the household head is Colombian. The
coefficient in front of treatment represents the impact for Ecuadorians, while the summation of
the coefficient in front of treatment and the coefficient in front of the interaction term represents
the impact for Colombians. As Table 6.12 reveals, the interaction term is not significant for the
food consumption measures. Thus, the impact of treatment on these measures for Colombians
is not significantly different to that of Ecuadorians. Across treatment arms, the differential effect
with respect to being Colombian is also not significant (Table 6.13).
Table 6.12 Differential impact on food consumption with respect to nationality, by pooled treatment
Log per capita food consumption Log per capita caloric intake
Pooled Treatment 0.15 0.11
(0.03)*** (0.03)***
Pooled Treatment X Colombian -0.06 -0.03
(0.07) (0.06)
Household head is Colombian -0.01 -0.02
(0.06) (0.06)
R2 0.35 0.31
N 1,985 2,006
Notes: Standard errors in parentheses clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. All
estimations contain baseline outcome variable, urban center fixed effects, and a full set of control variables.
Table 6.13 Differential impact on food consumption with respect to nationality, by treatment arms
Log per capita food consumption Log per capita caloric intake
Treatment==Food 0.18 0.15
(0.04)*** (0.04)***
Treatment==Cash 0.12 0.06
(0.04)*** (0.04)*
Treatment==Voucher 0.16 0.13
(0.04)*** (0.03)***
Food Treatment X Colombian -0.05 0.01
(0.08) (0.08)
Cash Treatment X Colombian -0.01 0.01
(0.08) (0.07)
Voucher Treatment X Colombian -0.11 -0.10
(0.07) (0.07)
Household head is Colombian -0.01 -0.02
(0.06) (0.06)
R2 0.36 0.32
N 1,985 2,006
Notes: Standard errors in parentheses clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. All
estimations contain baseline outcome variable, urban center fixed effects, and a full set of control variables.
.
44
Table 6.14 shows that both Ecuadorian and Colombian households benefit from the
pooled treatment in terms of increases in the dietary diversity measures, although Colombians
experience a significantly larger increase in the HDDS measure. Looking across treatment arms
(Table 6.15), we find that for Ecuadorians, vouchers have the largest impact on all three food
security measures; however, this is not true for the Colombians. For Colombians, vouchers have
the largest impact on the FCS measure, but food has the largest impact on HDDS and DDI. The
impact of food on HDDS is significantly larger for Colombian households compared to
Ecuadorian households. Furthermore, the impact of cash on HDDS is significantly larger for
Colombian households. For FCS and DDI, however, there is no differential impact with respect
to being Colombian for any treatment arm.
Table 6.14 Differential impact on dietary diversity with respect to nationality by pooled treatment
HDDS DDI FCS
Pooled treatment 0.32 2.32 7.39
(0.10)*** (0.41)*** (1.21)***
Pooled treatment X Colombian 0.50 0.57 0.58
(0.18)*** (0.68) (1.78)
Household head is Colombian -0.63 -1.44 -2.32
(0.16)*** (0.60)** (1.39)*
R2 0.19 0.29 0.18
N 2,087 2,087 2,087
Notes: Standard errors in parentheses clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. All estimations
contain baseline outcome variable, urban center fixed effects, and a full set of control variables.
Table 6.15 Differential impact on dietary diversity with respect to nationality, by treatment arms
HDDS DDI FCS
Treatment==Food 0.26 1.57 5.30
(0.12)** (0.53)*** (1.53)***
Treatment==Cash 0.25 2.19 6.38
(0.11)** (0.45)*** (1.46)***
Treatment==Voucher 0.43 2.87 9.51
(0.11)*** (0.46)*** (1.39)***
Food treatment X Colombian 0.84 1.38 2.78
(0.21)*** (0.88) (2.43)
Cash treatment X Colombian 0.53 0.68 0.19
(0.19)*** (0.74) (2.21)
Voucher treatment X Colombian 0.23 -0.05 -0.49
(0.22) (0.81) (2.68)
Household head is Colombian -0.63 -1.45 -2.39
(0.16)*** (0.60)** (1.40)*
R2 0.20 0.29 0.19
N 2,087 2,087 2,087
Notes: Standard errors in parentheses clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. All estimations
contain baseline outcome variable, urban center fixed effects, and a full set of control variables.
45
6.6 Impacts on Nonfood Expenditure
In this section we investigate whether the transfers are also used on nonfood items. Self-
reports indicate that for the most part, transfers are used on food consumption (Tables 5.10-
5.12), and indeed the impact results above confirm it. Nevertheless, we explore whether
nonfood expenditures also increase as a result of the program, and if so, whether it differs by
modality. We focus on the following nonfood expenditures that were collected at follow-up:
personal care (shampoo, toothpaste, etc.); household or kitchenware; devices or items for
communication; gas or electric; transportation; water or water treatment; housing;
entertainment; personal care outside of home (beauty salon, etc.); men’s clothes; women’s
clothes; children’s clothes; furniture or electronics; jewelry, toys and recreational goods;
education; and health. Similar to the tables above, we conduct ANCOVA estimates on pooled
treatment and by treatment arms. Given that the range in expenditures is not normally
distributed, we transform the dollar amount into logs. Thus the coefficients from the estimates
represent percent changes.
Table 6.16 reveals that across the 17 different nonfood expenditures, only expenditures
on toys marginally increases as a result of treatment. This reinforces the conclusion that the
transfers are mainly used on food and not on non-food items. Table 6.17 reveals that in general
the three transfer modalities are not used on non-food items, however, there are a few
significant increases at the 5 percent level. In particular, food leads to significant increases in
communication and men’s clothing. These increases however are small in monetary value, and
equivalent to a $1.15 and $1.6 increase respectively.
Table 6.18 explores whether the impacts on nonfood expenditure varies by nationality
and finds that with the exception of expenditures on household and kitchenware,
entertainment, and toys, there are no significant differences at the 5 percent level in impacts
across Colombians and Ecuadorians. For household and kitchenware, Colombians in the cash
arm show a significant increase of 19 percent and Colombians in the voucher arm show a
significant increase of 11 percent, and these are significantly different from the impact of cash
and vouchers for Ecuadorians. One possible explanation for this result is that Colombians, who
most likely just arrived to Ecuador, are using part of their cash to buy necessary household
items. In contrast, for entertainment and toys the impact for Colombians is significantly smaller
than that of Ecuadorians.
46
Table 6.16 Impact of pooled treatment on household expenditure (logs)
Personal
care
House
and
kitchen
appl.
Commun. Light
and
gas
Transp. Water Housing Entert. Men's
clothes
Women's
clothes
Child's
clothes
Furnit. and
electron.
Edu Health Services Jewelry Toys
Pooled
Treatment
0.12 0.03 0.09 -0.00 0.03 -0.01 0.12 0.02 0.12 0.01 0.07 0.01 0.14 0.09 0.01 0.01 0.03
(0.08) (0.03) (0.07) (0.06) (0.09) (0.06) (0.09) (0.03) (0.07) (0.07) (0.07) (0.04) (0.12) (0.12) (0.05) (0.01) (0.02)*
R2 0.12 0.02 0.24 0.21 0.20 0.24 0.43 0.02 0.06 0.06 0.18 0.02 0.45 0.09 0.04 0.01 0.05
N 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044
Notes: Standard errors in parentheses clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. All estimations contain baseline outcome variable, urban center fixed effects, and a
full set of control variables. Information on services, jewelry, and toys was not collected at baseline, and thus we do not control for it in the estimation.
Table 6.17 Impact of treatment arms on household expenditure (logs)
Personal
care
House
and
kitchen
appl.
Commun. Light
and
gas
Transp. Water Housing Entert. Men's
clothes
Women's
clothes
Child's
clothes
Furnit.
and
electron.
Edu Health Services Jewelry Toys
Treatment==Food 0.16 0.03 0.17 0.01 0.11 -0.02 0.14 0.03 0.22 0.12 0.06 0.04 0.01 0.04 0.06 0.02 0.03
(0.10) (0.03) (0.08)** (0.07) (0.12) (0.08) (0.11) (0.03) (0.10)** (0.08) (0.09) (0.05) (0.14) (0.16) (0.07) (0.02) (0.02)
Treatment==Cash 0.06 0.05 0.01 -0.00 -0.03 -0.03 0.12 0.00 0.05 -0.04 0.03 -0.03 0.10 0.12 0.00 -0.01 0.04
(0.08) (0.04) (0.09) (0.07) (0.10) (0.08) (0.11) (0.03) (0.08) (0.08) (0.08) (0.04) (0.14) (0.14) (0.06) (0.01) (0.02)*
Treatment==Voucher 0.16 0.01 0.10 -0.00 0.04 0.01 0.11 0.02 0.12 -0.01 0.11 0.02 0.26 0.09 -0.01 0.01 0.03
(0.09)* (0.03) (0.08) (0.06) (0.10) (0.07) (0.10) (0.03) (0.09) (0.08) (0.08) (0.05) (0.13)* (0.13) (0.06) (0.02) (0.02)
R2 0.12 0.02 0.24 0.21 0.21 0.24 0.43 0.02 0.06 0.06 0.18 0.02 0.45 0.09 0.04 0.01 0.05
N 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044
F test: Food=Voucher 0.00 0.40 0.94 0.03 0.51 0.16 0.09 0.27 1.38 2.84 0.44 0.14 3.29 0.15 1.38 0.39 0.05
P-value 0.99 0.53 0.33 0.86 0.48 0.69 0.77 0.60 0.24 0.09 0.51 0.71 0.07 0.70 0.24 0.53 0.83
F test: Cash=Voucher 1.78 1.25 1.59 0.00 0.83 0.44 0.01 0.40 0.83 0.33 1.16 1.39 1.21 0.07 0.09 2.41 0.43
P-value 0.18 0.27 0.21 0.96 0.36 0.51 0.93 0.53 0.36 0.56 0.28 0.24 0.27 0.80 0.76 0.12 0.51
F test: Food=Cash 1.54 0.31 3.56 0.05 2.10 0.03 0.04 1.19 4.56 5.35 0.11 2.30 0.48 0.35 0.90 3.39 0.12
P-value 0.22 0.58 0.06 0.82 0.15 0.86 0.84 0.28 0.03 0.02 0.74 0.13 0.49 0.55 0.34 0.07 0.72
Notes: Standard errors in parentheses clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. All estimations contain baseline outcome variable, urban center fixed effects, and a
full set of control variables. Information on services, jewelry, and toys was not collected at baseline, and thus we do not control for it in the estimation.
47
Table 6.18 Impact of treatment arms, by nationality, on household expenditure (logs)
Personal
care
House
and
kitchen
appl.
Commun. Light
and
gas
Transp. Water Housing Entert. Men's
clothes
Women's
clothes
Child's
clothes
Furnit.
and
electron.
Edu Health Services Jewelry Toys
Treatment==Food 0.18 -0.00 0.16 -0.02 0.19 -0.02 0.13 0.06 0.13 0.06 0.08 0.04 0.02 0.05 0.11 0.01 0.06
(0.11)* (0.04) (0.09)* (0.08) (0.11)* (0.09) (0.11) (0.03)* (0.11) (0.10) (0.10) (0.06) (0.17) (0.18) (0.08) (0.02) (0.03)**
Treatment==Cash 0.09 -0.01 -0.03 -0.01 -0.01 -0.03 0.06 0.03 0.00 -0.04 0.08 -0.01 0.04 0.19 0.00 -0.01 0.04
(0.09) (0.04) (0.10) (0.09) (0.10) (0.08) (0.11) (0.03) (0.09) (0.09) (0.09) (0.05) (0.16) (0.15) (0.06) (0.02) (0.02)
Treatment==Voucher 0.16 -0.04 0.13 -0.02 0.03 -0.01 0.05 0.03 0.08 -0.08 0.13 0.04 0.25 0.04 -0.02 0.01 0.05
(0.10) (0.04) (0.08) (0.08) (0.10) (0.07) (0.10) (0.03) (0.09) (0.09) (0.08) (0.06) (0.15)* (0.14) (0.07) (0.02) (0.02)*
Food Treatment X
Colombian
-0.07 0.09 0.03 0.09 -0.26 0.01 0.02 -0.10 0.29 0.18 -0.06 -0.02 -0.06 -0.04 -0.16 0.03 -0.09
(0.16) (0.08) (0.17) (0.14) (0.24) (0.13) (0.17) (0.05)** (0.15)* (0.15) (0.15) (0.08) (0.33) (0.27) (0.13) (0.05) (0.04)**
Cash Treatment X
Colombian
-0.12 0.20 0.15 0.02 -0.06 -0.03 0.20 -0.11 0.15 -0.03 -0.19 -0.06 0.27 -0.28 -0.00 0.02 0.01
(0.14) (0.10)** (0.14) (0.16) (0.19) (0.17) (0.22) (0.04)*** (0.12) (0.11) (0.13) (0.06) (0.26) (0.25) (0.12) (0.02) (0.04)
Voucher Treatment
X Colombian
0.01 0.15 -0.11 0.07 0.04 0.07 0.21 -0.04 0.10 0.29 -0.05 -0.09 0.01 0.21 0.03 -0.01 -0.08
(0.13) (0.07)** (0.13) (0.15) (0.18) (0.14) (0.20) (0.05) (0.12) (0.14)** (0.12) (0.08) (0.26) (0.25) (0.13) (0.02) (0.04)*
Household head is
Colombian
0.07 -0.09 -0.13 -0.20 -0.13 -0.07 0.33 0.07 -0.15 -0.19 -0.06 0.04 -0.17 -0.19 0.06 -0.02 0.03
(0.10) (0.06)* (0.10) (0.12) (0.15) (0.10) (0.13)** (0.03)** (0.09) (0.08)** (0.09) (0.05) (0.19) (0.18) (0.10) (0.02) (0.03)
R2 0.12 0.02 0.24 0.21 0.21 0.24 0.43 0.02 0.07 0.06 0.18 0.02 0.45 0.09 0.04 0.01 0.05
N 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044 2,044
Notes: Standard errors in parentheses clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. All estimations contain baseline outcome variable, urban center fixed effects, and a
full set of control variables. Information on services, jewelry, and toys was not collected at baseline, and thus we do not control for it in the estimation.
48
7. Impact on Social Capital: Trust, Discrimination, and Participation
7.1 Baseline and Follow-Up Descriptive Statistics
Although the program did not have explicit mandates to change social capital, because
of the mixed nationality targeting and because of the interaction of beneficiaries during training
sessions, it was hypothesized that the program could change certain components of social
capital. Social capital is defined by Putnam (2000) as ‘‘features of social life, networks, norms,
trust that enable participants to act together more effectively to pursue shared objectives.’’ We
focus on the social capital components of trust, discrimination, and community participation in
our impact analysis. Questions on discrimination and community participation in the
household surveys are asked at the household level (“has anyone in the household. . . “),
however, questions on trust are at the individual level; and thus, our analysis is at the
individual level. Given that we want to observe social capital as measured by the same
individuals at baseline and follow-up, we restrict the sample to 1,879 of the 2,122 households
where the same individuals were administered the survey between rounds.
Similar to Alesina and Ferrara (2002) and Labonne and Chase (2011), we divide trust
into trust of individuals and trust/confidence in institutions. The questions on trust are on a 4-
point scale—disagree, somewhat disagree, somewhat agree, and agree. We create an indicator
for trust of individuals that equals 1 if respondent agrees with any one of the following
questions: “I can trust the majority of people,” “I can trust my neighbor to send an important
letter,” and “I can trust my neighbor to look after my house while I am away.” Similarly, we
create an indicator for trust/confidence in institutions that equals 1 if respondent agrees with
any of the following questions: “The government would help my family if I had an emergency,”
“Politicians represent my interests,” and “I can go to the police for help if I am a victim of
crime.” As Alesina and Ferrara point out, a respondent may answer affirmatively to the
question about trusting others, even though in his or her actual behavior they may not exhibit
trusting behavior or actions. Consequently, to account for this potential ambiguity, we follow
Alesina and Ferrara and categorize as trusting only those who “agree” with the questions and
not those who “somewhat agree.”
Questions on discrimination and participation ask if anyone in the household has felt
discriminated against or participated in groups in the last 6 months. The survey asks about 10
different types of discrimination a household could have experienced: discrimination due to
color or ethnic origin, gender, lack of money or socioeconomic status, occupation, political
views, sickness or disability, nationality, religion, physical appearance, and other causes. We
create an indicator for discrimination that equals 1 if the respondent answers yes to anyone in
the household having experienced any of the 10 different types of discrimination. For
community participation, the survey asks whether or not anyone in the household has
participated in the following five groups or associations: agriculture or business association,
union, or cooperative; religious or spiritual group; community or neighborhood group; political
movement or group; other groups such as NGOs, education, or cultural group. We create an
indicator for community participation that equals 1 if a respondent answers yes to participating
in any of the five groups.
49
Tables 7.1 and 7.2 show the percent of individuals that trust other individuals, trust
institutions, experience discrimination, and participate in groups or associations at baseline and
follow-up. At baseline, 60 percent of respondents trust other individuals, while 87 percent trust
institutions. Interestingly, Colombians trust individuals (67 percent compared to 57 percent)
and institutions (89 percent versus 86 percent) significantly more than Ecuadorians.
Approximately 41 percent of respondents had experienced discrimination in the last 6 months,
with Colombians experiencing significantly more discrimination as compared to Ecuadorians
(49 percent versus to 36 percent). Approximately 49 percent of respondents had participated in
a community group in the last 6 months, with Colombians participating significantly less than
Ecuadorians (46 percent compared to 51 percent). Across treatment and comparison arms there
are significant differences in means at baseline for the indicators of trust in institutions and
participation, thus controlling for baseline values is critical for this analysis. At follow-up, the
percent of respondents trusting other individuals and institutions increases slightly, as does the
percent participating in groups. For trust in individuals, the percentage in the comparison arm
increases slightly, while the percent in the treatment arm remains constant. For trust of
institutions and participation in groups, the percent in the comparison arm decreases, while the
percent in the treatment arm increases. The percent of individuals experiencing discrimination
decreases to 34 percent at follow-up. Both treatment and comparison arms experience decreases
in discrimination; however, the decrease for the treatment arm is larger in magnitude.
Consequently, significantly more individuals in the treatment arm trust institutions at follow-
up and significantly less experience discrimination.
Table 7.1 Baseline means, by pooled treatment
All Comparison Treatment Difference
Trust in individuals 0.60 0.59 0.61 0.02
(0.01) (0.02) (0.01) ( 0.03)
Trust in institutions 0.87 0.90 0.86 -0.04
(0.01) (0.01) (0.01) ( 0.02)**
Discrimination 0.41 0.40 0.41 0.00
(0.01) (0.02) (0.01) ( 0.03)
Participation 0.49 0.53 0.48 -0.06
(0.01) (0.02) (0.01) ( 0.03)**
N 1,879 506 1,373
Notes: Standard errors in parentheses. * p < 0.1 ** p < 0.05; *** p < 0.01.
Table 7.2 Follow-up means, by pooled treatment
All Comparison Treatment Difference
Trust in individuals 0.61 0.62 0.61 -0.01
(0.01) (0.02) (0.01) ( 0.03)
Trust in institutions 0.90 0.86 0.91 0.06
(0.01) (0.02) (0.01) ( 0.02)***
Discrimination 0.34 0.38 0.32 -0.06
(0.01) (0.02) (0.01) ( 0.03)**
Participation 0.52 0.50 0.53 0.04
(0.01) (0.02) (0.01) ( 0.03)
N 1,878 505 1,373
Notes: Standard errors in parentheses. * p < 0.1 ** p < 0.05; *** p < 0.01.
50
7.2 Impacts on Trust, Discrimination, and Community Participation
Tables 7.3–7.4 present the ANCOVA estimates of being in the treatment arm on social
capital measures. In Table 7.3, we combine all three treatment arms (food, cash, and voucher)
and estimate the impact of pooled treatment on trust of individuals, trust of institutions,
discrimination, and participation in groups. We present the results with and without controlling
for any covariates. Given that the analysis is at the individual level, covariates for education,
age, sex, nationality, and marital status are at the level of the individual and not household
head. We focus on columns 2, 4, 6, and 8 of Table 7.3 which control for the full set of covariates.
Pooled treatment significantly decreases discrimination by 6 percentage points and increases
participation in groups by 6 percentage points. These finding demonstrate the success of the
program in bringing individuals together from different backgrounds. Other factors that
contribute significantly to the social capital measures are wealth, age, and being Colombian. In
particular, being Colombian and older is significantly associated with more discrimination,
while being in the top wealth quintile is associated with less discrimination and trust of
individuals.
Table 7.4 presents the ANCOVA estimates for each treatment arm separately, and
conducts Wald tests to exam whether the estimates from each treatment arm are significantly
different from each other. We conduct the estimates using a full set of covariates; however, we
report only treatment coefficients for simplicity. Even though all three modalities decrease
discrimination and the size of the coefficient is similar across modalities, only the cash arm
leads to a significant impact. Similarly, only the cash arm leads to a significant decrease in trust
of individuals and this decrease is significantly different from the impact of the food arm.
Although cash leads to significant decreases in trust of individuals, it significantly increases
trust of institutions. For participation in groups, only the voucher arm leads to a significant
increase, and this increase is significantly different from the increase of the cash arm.
Table 7.4 reveals the percentage point change of each social capital indicator that is due
to specific treatment modalities. We convert the size of the change to percent changes from
baseline means. Figure 7.1 reveals that the percentage decrease in discrimination is large,
especially for the cash arm: 19.5 percent decrease in discrimination for the cash arm compared
to 12.2 percent for the food arm and 9.8 percent for the voucher arm. Cash also leads to a large
decrease of 11.7 percent in trust of individuals. For participation in groups or associations, only
the voucher arm experiences a large and significant increase of 20.4 percent.
51
Table 7.3 Impact of pooled treatment on social capital measures
Trust Individual Trust Institution Discrimination Participation
(1) (2) (3) (4) (5) (6) (7) (8)
Pooled treatment -0.01 -0.04 0.06 0.04 -0.06 -0.06 0.05 0.06
(0.03) (0.03) (0.03)** (0.02) (0.03)* (0.03)* (0.03) (0.03)*
Colombian -0.03 -0.00 0.08 -0.01
(0.03) (0.02) (0.03)*** (0.03)
Female -0.05 0.01 -0.01 0.02
(0.03) (0.02) (0.03) (0.03)
Age 0.00 -0.00 0.00 0.00
(0.00) (0.00) (0.00)*** (0.00)***
Secondary education or higher 0.02 0.00 0.02 -0.01
(0.03) (0.02) (0.03) (0.03)
Married -0.02 -0.00 -0.03 0.03
(0.03) (0.02) (0.03) (0.03)
Household size 0.01 0.00 0.00 0.01
(0.01) (0.01) (0.01) (0.01)
Number of children 0-5 years -0.03 -0.03 0.02 -0.00
(0.02) (0.01)* (0.02) (0.02)
Number of children 6-15 years -0.00 0.01 0.02 0.01
(0.01) (0.01) (0.02) (0.02)
Wealth index: 2nd quintile -0.04 0.02 -0.03 0.00
(0.03) (0.02) (0.04) (0.04)
Wealth index: 3rd quintile -0.02 0.00 -0.06 0.02
(0.04) (0.02) (0.04) (0.04)
Wealth index: 4th quintile 0.00 0.01 -0.05 0.05
(0.04) (0.02) (0.04) (0.04)
Wealth index: 5th quintile -0.11 -0.01 -0.07 0.06
(0.04)*** (0.02) (0.03)** (0.04)
Baseline trust in individuals 0.17 0.17
(0.03)*** (0.03)***
Baseline trust in institutions 0.06 0.06
(0.02)** (0.02)**
Baseline discrimination 0.25 0.23
(0.02)*** (0.02)***
Baseline participation 0.16 0.14
(0.03)*** (0.03)***
Constant 0.52 0.60 0.80 0.86 0.28 0.14 0.41 0.21
(0.03)*** (0.08)*** (0.03)*** (0.05)*** (0.03)*** (0.07)** (0.02)*** (0.07)***
R2 0.03 0.04 0.01 0.03 0.07 0.09 0.03 0.04
N 1,878 1,878 1,878 1,878 1,879 1,879 1,879 1,879
Notes: Standard errors in parentheses clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. Columns 2, 4, 6,
and 8 contain urban center fixed effects.
52
Table 7.4 Impact of treatment modalities on social capital measures
Trust
Individual
Trust
Institution Discrimination Participation
Treatment==Food 0.03 0.02 -0.05 0.06
(0.04) (0.03) (0.04) (0.05)
Treatment==Cash -0.07 0.06 -0.08 0.01
(0.04)* (0.03)** (0.04)** (0.04)
Treatment==Voucher -0.05 0.03 -0.04 0.10
(0.04) (0.03) (0.04) (0.04)**
R2 0.05 0.04 0.09 0.05
N 1,878 1,878 1,879 1,879
F test: Food=Voucher 4.60 0.07 0.11 0.59
P-value 0.03 0.80 0.74 0.44
F test: Cash=Voucher 0.11 2.99 1.29 4.75
P-value 0.74 0.09 0.26 0.03
F test: Food=Cash 5.70 2.57 0.45 1.16
P-value 0.02 0.11 0.50 0.28
Notes: Standard errors in parentheses clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. All estimations
contain baseline outcome variable, urban center fixed effects, and a full set of control variables.
Figure 7.1 Percent change in social capital indicators, by treatment arms
Note: * p < 0.1 ** p < 0.05; *** p < 0.01.
7.3 Impacts on Individual Questions
The summary indicators on trust, discrimination, and community participation hide
some interesting patterns of the individual questions, and thus in this section we investigate the
impact of treatment on each individual question that makes up the summary indicators. Similar
to previous tables, we present ANCOVA estimates. Closer analysis of trust indicators reveals
5.0 2.3
-12.2
12.2
-11.7*
6.9**
-19.5**
2.0
-8.3
3.4
-9.8
20.4**
-30.0
-20.0
-10.0
0.0
10.0
20.0
30.0
Trust Individual Trust Institutions Discrimination Participation%
Food
Cash
Voucher
53
that the cash and voucher arms lead to a significant decrease in beneficiaries’ trust in neighbors
to take care of one’s home, and the size of the coefficient for both arms is significantly different
to that of the food arm (Table 7.5). The cash arm also significantly decreases trust in neighbors
to send a letter, and again the size of the coefficient is significantly different to that of the food
arm. Although cash decreases beneficiaries’ trust in neighbors, it also significantly increases
trust/confidence in the police.
Table 7.5 Impact of treatment modalities on trust indicators
Majority of
people
Neighbor to
send letter
Neighbor to
take care of
house while
away
Government
to help my
family in
emergency
Politicians
to represent
my interests
Police if I
am a victim
of crime
Treatment==Food 0.03 0.02 0.02 0.05 -0.03 0.03
(0.04) (0.04) (0.04) (0.04) (0.05) (0.03)
Treatment==Cash 0.00 -0.07 -0.10 0.05 0.01 0.07
(0.03) (0.04)* (0.04)** (0.04) (0.04) (0.03)***
Treatment==Voucher -0.02 -0.05 -0.07 -0.00 -0.04 0.02
(0.03) (0.03) (0.04)* (0.04) (0.04) (0.03)
R2 0.05 0.05 0.06 0.04 0.03 0.03
N 1,878 1,878 1,878 1,878 1,878 1,878
F test: Food=Voucher 1.34 3.86 4.31 1.61 0.06 0.08
P-value 0.25 0.05 0.04 0.21 0.80 0.78
F test: Cash=Voucher 0.30 0.30 0.69 1.45 1.26 5.51
P-value 0.58 0.59 0.41 0.23 0.26 0.02
F test: Food=Cash 0.48 5.28 8.52 0.00 0.55 2.27
P-value 0.49 0.02 0.00 0.98 0.46 0.13
Notes: Standard errors in parentheses clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. All estimations
contain baseline outcome variable, urban center fixed effects, and a full set of control variables.
Table 7.6 shows the impact of treatment arms on individual indicators of discrimination,
and reveals that both the cash and voucher lead to decreases of 3 percentage points in
discrimination due to socioeconomic status and occupation, however, this decrease is not
significant at the 10 percent level. All three treatment arms lead to increases in “other” types of
discrimination, which could reflect negative stigma attached to receiving aid.
Table 7.7 shows the impact of treatment arms on individual indicators of community
participation, and reveals that the increase in participation is due solely to increases in
participation in NGOs, or education or cultural groups. This is not surprising, given that
participants in the program most likely see themselves as participating in an NGO. On the other
hand, the cash arm also leads to a significant decrease in participation of community
associations.
54
Table 7.6 Impact of treatment modalities on discrimination indicators
Color or
race Gender
Lack of money
or socio-
economic status Occupation
Political
views
Sickness or
disability Nationality Religion
Physical
appearance Other
Treatment==Food 0.02 0.00 0.00 0.00 0.01 0.01 -0.01 0.00 0.03 0.02
(0.02) (0.02) (0.04) (0.03) (0.02) (0.02) (0.03) (0.02) (0.02)* (0.01)*
Treatment==Cash 0.02 0.00 -0.03 -0.03 -0.01 -0.01 -0.02 -0.01 0.02 0.02
(0.01) (0.02) (0.03) (0.03) (0.01) (0.02) (0.02) (0.02) (0.01) (0.01)**
Treatment==Voucher 0.01 -0.02 -0.03 -0.03 -0.01 -0.02 0.01 -0.01 0.01 0.02
(0.02) (0.02) (0.03) (0.02) (0.01) (0.01)* (0.02) (0.01) (0.01) (0.01)**
R2 0.06 0.04 0.07 0.07 0.04 0.07 0.19 0.07 0.05 0.02
N 1,879 1,879 1,879 1,879 1,879 1,879 1,879 1,879 1,879 1,879
F test: Food=Voucher 0.33 1.74 1.03 1.44 1.63 4.25 0.60 0.42 1.74 0.01
P-value 0.57 0.19 0.31 0.23 0.20 0.04 0.44 0.52 0.19 0.92
F test: Cash=Voucher 0.06 1.17 0.01 0.02 0.08 1.05 1.46 0.01 0.96 0.05
P-value 0.80 0.28 0.93 0.88 0.78 0.31 0.23 0.93 0.33 0.81
F test: Food=Cash 0.14 0.05 0.95 1.62 2.26 1.28 0.07 0.29 0.28 0.01
P-value 0.71 0.83 0.33 0.21 0.14 0.26 0.79 0.59 0.60 0.91
Notes: Standard errors in parentheses clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. All estimations contain baseline outcome variable, urban center
fixed effects, and a full set of control variables.
55
Table 7.7 Impact of treatment modalities on community participation
Agricultural or
business associations
or cooperatives or
unions
Religious or
spiritual
group
Community
association
Political
group
NGO or
education or
cultural
groups
Treatment==Food -0.02 -0.03 -0.00 0.01 0.17
(0.02) (0.04) (0.04) (0.01) (0.05)***
Treatment==Cash -0.02 0.01 -0.06 -0.00 0.09
(0.02) (0.04) (0.03)* (0.01) (0.04)**
Treatment==Voucher -0.02 0.00 0.05 -0.01 0.13
(0.02) (0.04) (0.04) (0.01) (0.03)***
R2 0.07 0.09 0.06 0.03 0.05
N 1,879 1,879 1,879 1,879 1,879
F test: Food=Voucher 0.00 0.67 1.13 2.84 0.51
P-value 0.95 0.41 0.29 0.09 0.48
F test: Cash=Voucher 0.00 0.08 9.05 1.91 0.91
P-value 1.00 0.78 0.00 0.17 0.34
F test: Food=Cash 0.00 1.20 2.31 0.81 1.96
P-value 0.95 0.28 0.13 0.37 0.16
Notes: Standard errors in parentheses clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. All estimations
contain baseline outcome variable, urban center fixed effects, and a full set of control variables.
7.4 Impacts by Nationality
In Tables 7.8 and 7.9 we explore whether the impact of treatment differed by Ecuadorian
and Colombian participants. We create an interaction term of treatment with an indicator for
whether or not an individual is Colombian, and the coefficient in front of this interaction term
represents the differential impact of treatment. In both tables, we conduct the analysis with the
full set of individual, household, and urban center controls; however, we present only the
coefficients on treatment, being Colombian, and the interaction of the two indicators. As Table
7.8 reveals, the differential impact on trust and discrimination is not significant. However, for
participation the impact is significantly larger for Colombians. In particular, treatment leads to a
2 percentage point increase in participation for Ecuadorians while it leads to an 11 percentage
point increase for Colombians. Looking across treatment arms (Table 7.9), we again only find a
significant differential impact with respect to being Colombian for the participation indicator. In
particular, the impact of cash on participation is significantly larger for Colombians than for
Ecuadorians.
56
Table 7.8 Differential impact with respect to nationality, by pooled treatment
Trust Individual Trust Institution Discrimination Participation
Pooled treatment -0.02 0.02 -0.06 0.02
(0.04) (0.02) (0.04) (0.04)
Pooled treatment X Colombian -0.04 0.04 -0.01 0.09
(0.06) (0.05) (0.06) (0.05)*
Colombian -0.00 -0.03 0.09 -0.08
(0.05) (0.05) (0.06) (0.04)*
R2 0.04 0.03 0.09 0.05
N 1,878 1,878 1,879 1,879
Notes: Standard errors in parentheses clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. All estimations
contain baseline outcome variable, urban center fixed effects, and a full set of control variables.
Table 7.9 Differential impact with respect to nationality, by treatment arms
Trust Individual Trust Institution Discrimination Participation
Treatment==Food 0.06 0.00 -0.04 0.05
(0.05) (0.03) (0.05) (0.06)
Treatment==Cash -0.07 0.05 -0.08 -0.04
(0.05) (0.03)** (0.05) (0.05)
Treatment==Voucher -0.03 0.01 -0.05 0.06
(0.04) (0.03) (0.04) (0.04)
Food Treatment X Colombian -0.07 0.04 -0.06 0.02
(0.07) (0.05) (0.08) (0.07)
Cash Treatment X Colombian 0.01 0.03 -0.00 0.15
(0.07) (0.05) (0.07) (0.07)**
Voucher Treatment X Colombian -0.06 0.04 0.02 0.10
(0.07) (0.05) (0.07) (0.06)
Colombian 0.00 -0.03 0.09 -0.08
(0.05) (0.05) (0.06) (0.04)*
R2 0.05 0.04 0.09 0.05
N 1,878 1,878 1,879 1,879
Notes: Standard errors in parentheses clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. All estimations
contain baseline outcome variable, urban center fixed effects, and a full set of control variables.
57
8. Anemia
8.1 Indicators and Descriptive Statistics
Hemoglobin biomarkers were collected for all children aged 6 to 59 months and for
female adolescents aged 10 to 16 residing in surveyed households. Anemia cut-off
classifications are calculated according to the WHO recommendations (2011) by age group
(Table 8.1) and adjusted for pregnancy status and altitude using the recommendations of
Sullivan and colleagues (Sullivan et al. 2008). Due to the very low percentages of children or
adolescents with severe anemia in the sample, we combine severe and moderate classification in
the analysis.
Table 8.1 Anemia classifications
Anemia status
Age Severe Moderate Mild
6-59 months hb < 7 7 ≤ hb < 10 10 ≤ hb < 11.0
5-6 years and 10-11 years hb < 8 8 ≤ hb < 11 11 ≤ hb < 11.5
12-17 years hb < 8 8 ≤ hb < 11 11 ≤ hb < 12.0
In total there are 785 children 6 to 59 months and 432 adolescents 10 to 16 years old that
had their hemoglobin measured at baseline and follow-up. Sample sizes reflect both changes in
household composition as well as availability of children to be tested in both rounds at the time
of the interview. Attrition rates for the anemia sample are approximately 20 percent in both the
children and adolescent sample. Conducting t-tests reveals that there are significant differences
in terms of female headship, ethnicity of household head and wealth among the attrited sample
of children aged 6 to 59 months, and significant differences in terms of ethnicity of household
head and wealth among the attrited sample of adolescents. There is no differential attrition by
pooled treatment status in either the children aged 6 to 59 month sample or the adolescent
sample.
Table 8.2 presents anemia levels for children between 6 and 59 months at baseline by
pooled treatment. Average hemoglobin levels are 10.85 g/dl in the full sample. Overall levels of
anemia are high (48 percent); a little less than half is concentrated in the mild (21 percent) level,
while a little more than half (26 percent) is concentrated in the moderate to severe levels. There
are no significant differences between comparison and treatment at baseline, indicating success
of the randomization.
Table 8.3 presents anemia levels for children between 6 and 59 months at follow-up by
pooled treatment. Overall hemoglobin levels have increased approximately 0.4 g/dl to 11.23,
which may be a function of children aging over the panel period. The percentage with any
anemia decreases to 41 percent, and this decrease is entirely due to the decrease in moderate
and severe anemia from 26 percent to 18 percent respectively. Similar to the baseline, there are
no significant differences between treatment and comparison groups.
58
Table 8.2 Baseline anemia measures among 6–59 month old children, by pooled treatment
All Comparison Treatment Difference
Hemoglobin level (g/dl) 10.85 10.97 10.81 0.16
(0.05) (0.11) (0.06) ( 0.13)
Any anemia indicator 0.48 0.44 0.49 -0.05
(0.02) (0.04) (0.02) ( 0.04)
Mild anemia indicator 0.21 0.18 0.22 -0.04
(0.01) (0.03) (0.02) ( 0.03)
Moderate/severe anemia indicator 0.26 0.26 0.27 -0.01
(0.02) (0.03) (0.02) ( 0.04)
N 785 197 588
Note: Standard errors in parenthesis. * p < 0.1 ** p < 0.05; *** p < 0.01.
Table 8.3 Follow-up anemia measures among 6–59 month old children, by pooled treatment
All Comparison Treatment Difference
Hemoglobin level (g/dl) 11.23 11.35 11.18 0.17
(0.05) (0.09) (0.05) ( 0.11)
Any anemia indicator 0.41 0.37 0.43 -0.06
(0.02) (0.03) (0.02) ( 0.04)
Mild anemia indicator 0.23 0.22 0.24 -0.02
(0.02) (0.03) (0.02) ( 0.03)
Moderate/severe anemia indicator 0.18 0.15 0.19 -0.04
(0.01) (0.03) (0.02) ( 0.03)
N 785 197 588
Note: Standard errors in parenthesis. * p < 0.1 ** p < 0.05; *** p < 0.01.
Table 8.4 shows results of baseline anemia measures for the sample of adolescent girls
by pooled treatment. Overall levels of anemia are much lower than for the sample aged 6 to 59
months, with 23 percent of adolescents having any form of anemia, which is split between those
with mild anemia at 16 percent and those with moderate to severe anemia at 8 percent.
Hemoglobin levels are 12.55 g/dl for the entire sample and there are no significant differences
between comparison and treatment arms, indicating the success of the randomization.
Table 8.4 Baseline anemia measures among adolescents aged 10 – 16, by pooled treatment
All Comparison Treatment Difference
Hemoglobin level (g/dl) 12.55 12.53 12.56 -0.03
(0.06) (0.10) (0.07) (0.12)
Any Anemia Indicator 0.23 0.24 0.23 0.01
(0.02) (0.04) (0.02) (0.04)
Mild Anemia Indicator 0.16 0.15 0.16 -0.00
(0.02) (0.03) (0.02) (0.04)
Moderate/severe Anemia Indicator 0.08 0.09 0.07 0.02
(0.01) (0.02) (0.01) (0.03)
N 432 136 296
Note: Standard errors in parenthesis. * p < 0.1 ** p < 0.05; *** p < 0.01.
Table 8.5 presents results of follow-up anemia measures for the sample of adolescent
girls by pooled treatment. Again, there is little movement in overall hemoglobin levels,
59
however, for the adolescent girls, there is a slight upward trend in any anemia (28 percent
versus 23 percent in the baseline), and these slight increases are seen in both mild and moderate
measures. Finally, there are no significant differences between treatment and comparison
groups.
Table 8.5 Follow-up anemia measures among adolescents aged 10 – 16, by pooled treatment
All Comparison Treatment Difference
Hemoglobin level (g/dl) 12.46 12.37 12.51 -0.14
(0.06) (0.11) (0.07) ( 0.13)
Any anemia indicator 0.28 0.29 0.28 0.00
(0.02) (0.04) (0.03) ( 0.05)
Mild anemia indicator 0.18 0.15 0.20 -0.04
(0.02) (0.03) (0.02) ( 0.04)
Moderate/severe anemia indicator 0.10 0.14 0.09 0.05
(0.01) (0.03) (0.02) ( 0.03)
N 432 136 296
Note: Standard errors in parenthesis. * p < 0.1 ** p < 0.05; *** p < 0.01.
8.2 Impact of Treatment on Anemia Indicators for Children 6-59 Months Old
In Tables 8.6 and 8.7 we combine all three treatment arms (food, cash, and voucher) and
estimate the impact of pooled treatment on hemoglobin levels and indicators of anemia. We
present the ANCOVA estimates with and without controlling for additional covariates. We use
the same modeling as in previous analysis; however we include additional child-specific control
variables, age indicators (in 6 month splines for the under 5 group and in year indicators for the
adolescent group), sex of child, indicator for whether or not they received an iron dose in the
last 6 months and if the girl is menstruating at the time of the survey for the adolescent girl
analysis.3 Table 8.6 shows that there is virtually no impact found of the pooled treatment on any
of the anemia measures. This result is robust to the inclusion of control variables, although
coefficients change in magnitude, none are statistically significant (Table 8.7).
Table 8.6 Impact of pooled treatment on anemia measures among 6–59 month old children
Hemoglobin (g/dl) Any anemia Moderate or severe
Pooled treatment -0.11 0.05 0.04
(0.11) (0.04) (0.03)
Baseline anemia indicator 0.35 0.21 0.14
(0.03)*** (0.04)*** (0.03)***
Constant 7.49 0.27 0.11
(0.36)*** (0.03)*** (0.02)***
R2 0.16 0.05 0.03
N 785 785 785
Note: Standard errors in parenthesis clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01.
3 We also conduct robustness checks with alternate functional forms for age (continuous month and year
variables and indicators), and our results are robust to differences in functional form.
60
Table 8.7 Impact of pooled treatment on anemia measures among 6–59 month old children with
covariates
Hemoglobin (g/dl) Any anemia Moderate or severe
Pooled treatment -0.14 0.05 0.06
(0.13) (0.05) (0.03)
Child is male 0.12 -0.06 -0.04
(0.09) (0.03)* (0.03)
Received Iron Supplement in last 6 months -0.03 0.00 -0.02
(0.10) (0.04) (0.03)
Household head is female 0.08 -0.02 -0.03
(0.11) (0.04) (0.03)
Age of household head -0.00 0.00 0.00
(0.00) (0.00) (0.00)
Household head is Colombian 0.02 -0.06 -0.02
(0.09) (0.04) (0.03)
Household head has at least secondary education 0.03 0.03 -0.01
(0.11) (0.04) (0.03)
Household size -0.01 0.00 0.01
(0.02) (0.01) (0.01)
Wealth index: 2nd quintile -0.22 0.03 0.10
(0.17) (0.06) (0.04)**
Wealth index: 3rd quintile -0.01 0.01 -0.01
(0.15) (0.06) (0.04)
Wealth index: 4th quintile -0.20 0.04 0.02
(0.13) (0.06) (0.04)
Wealth index: 5th quintile -0.12 -0.02 0.05
(0.17) (0.06) (0.05)
Baseline anemia measure 0.30 0.17 0.09
(0.03)*** (0.04)*** (0.03)***
Constant 8.13 0.43 0.19
(0.41)*** (0.12)*** (0.09)**
R2 0.22 0.11 0.10
N 785 785 785
Notes: Standard errors in parenthesis clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. All estimations
contain child's age using fixed effects in six month age group splines and urban center fixed effects.
Table 8.8 replicates the ANCOVA estimates for hemoglobin levels and anemia measures
for each treatment arm separately, and reports Wald tests examining whether the estimates
from each treatment arm are significantly different from each other. Specifications include a full
set of control variables; however, for simplicity we only present the coefficients from the
different treatment arms. Table 8.8 shows that there is a significant decrease in hemoglobin and
increase in any anemia from the food treatment arm, and this increase in anemia is significantly
different to that of the voucher arm. Cash also leads to weakly significant increases in moderate
to severe anemia. When we consider interactions with ethnicity, we can see that the impact of
food on anemia is being driven entirely by increases in the moderate or severe indicator for the
Colombian sample (Table 8.9).
61
Table 8.8 Impact of treatment modalities on anemia measures among 6–59 month old children with
covariates
Hemoglobin (g/dl) Any anemia Moderate or severe
Treatment==Food -0.25 0.12 0.06
(0.15)* (0.06)** (0.04)
Treatment==Cash -0.13 0.04 0.09
(0.14) (0.05) (0.04)*
Treatment==Voucher -0.06 0.00 0.02
(0.15) (0.05) (0.04)
R2 0.22 0.11 0.10
N 785 785 785
F test: Food=Voucher 2.09 4.98 1.06
P-value 0.15 0.03 0.30
F test: Cash=Voucher 0.35 0.69 2.07
P-value 0.55 0.41 0.15
F test: Food=Cash 0.94 2.60 0.34
P-value 0.34 0.11 0.56
Notes: Standard errors in parenthesis clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. All estimations
control for child's age using fixed effects in six month age group splines, child's sex, indicator of receiving an iron
supplementation in the last 6 months, ethnicity, sex and education of household head, household size, number of
children, wealth quintiles, baseline outcome variable, and contain urban center fixed effects.
Table 8.9 Differential impact on anemia measures among 6 - 59 month old children with covariates
with respect to ethnicity, by treatment arms
Hemoglobin (g/dl) Any anemia Moderate or severe
Treatment==Food -0.13 0.08 0.02
(0.17) (0.06) (0.05)
Treatment==Cash -0.19 0.08 0.10
(0.15) (0.05) (0.05)**
Treatment==Voucher -0.03 0.01 0.01
(0.17) (0.06) (0.04)
Food treatment X Colombian -0.52 0.19 0.21
(0.23)** (0.11)* (0.08)***
Cash treatment X Colombian 0.34 -0.18 -0.12
(0.21) (0.11) (0.08)
Voucher treatment X Colombian -0.14 -0.02 0.04
(0.25) (0.10) (0.08)
Household head is Colombian 0.10 -0.05 -0.04
(0.16) (0.08) (0.05)
R2 0.23 0.12 0.11
N 785 785 785
Notes: Standard errors in parenthesis clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. All estimations
control for child's age using fixed effects in six month age group splines, child's sex, indicator of receiving an iron
supplementation in the last 6 months, ethnicity, sex and education of household head, household size, number of children,
wealth quintiles, baseline outcome variable, and contain urban center fixed effects.
62
8.3 Impact of Treatment on Anemia Indicators for Adolescent Females
In Tables 8.10 and 8.11 we combine all three treatment arms (food, cash, and voucher)
and estimate the impact of pooled treatment on hemoglobin levels and indicators of anemia for
adolescent females using the same modeling as for the sample aged 6 to 59 months. We present
the ANCOVA estimates with and without controlling for additional covariates. As Tables 8.10
and 8.11 reveal, we find no significant impact of the combined treatment and this is true
whether or not we control for covariates.
Table 8.10 Impact of pooled treatment on anemia measures among adolescent girls aged 10–16
Hemoglobin (g/dl) Any anemia Moderate or severe
Pooled treatment 0.13 -0.00 -0.05
(0.13) (0.05) (0.03)
Baseline anemia indicator 0.26 0.12 0.05
(0.06)*** (0.07)* (0.08)
Constant 9.16 0.26 0.14
(0.76)*** (0.04)*** (0.03)***
R2 0.06 0.01 0.01
N 432 432 432
Note: Standard errors in parenthesis clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01.
Table 8.12 reports the ANCOVA estimates for hemoglobin and anemia measures among
the adolescent girls sample for each treatment arm separately, and conducts Wald tests to
examine whether the estimates from each treatment arm are significantly different from each
other. Specifications include a full set of control variables; however, for simplicity we only
present the coefficients from the different treatment arms. Table 8.12 shows no significant
impact of any treatment arm across anemia measures, and this result holds when we consider
interactions with Colombian indicators (Table 8.13).
Taken together the results from anemia can be summarized in the following way. First,
although levels of anemia are high, particularly for the younger age group, the intervention did
not have a positive impact on anemia reduction for either children aged 6 to 59 months, or for
adolescent girls aged 10 to 16 years. This result could be explained by the relatively short
intervention period. In addition, the focus of the intervention was on food security and
nutrition in general, and not focused specifically on anemia reduction. Based on the results, it is
possible, and is corroborated by household food group caloric results, that for young children,
consumption moved away from more diverse diets and towards consumption of staples that
were part of the food transfer for that treatment group. Further analysis of the anemia data will
include explicit methods to account for attrition between panel periods.
63
Table 8.11 Impact of pooled treatment on anemia measures among adolescent girls aged 10–16 with
covariates
Hemoglobin (g/dl) Any anemia Moderate or severe
Pooled treatment -0.02 0.04 -0.03
(0.11) (0.04) (0.03)
Received iron supplement in last 6 months -0.04 -0.01 0.01
(0.17) (0.06) (0.05)
Currently menstruating -0.10 0.06 -0.01
(0.17) (0.07) (0.05)
Household head is female -0.08 0.08 0.02
(0.14) (0.05) (0.04)
Age of household head 0.00 0.00 0.00
(0.00) (0.00) (0.00)
Household head is Colombian 0.24 -0.11 -0.07
(0.13)* (0.05)** (0.03)**
Household head has at least secondary education 0.15 -0.05 -0.05
(0.14) (0.05) (0.04)
Household size -0.04 0.02 0.01
(0.03) (0.01) (0.01)
Wealth index: 2nd quintile 0.25 -0.06 -0.05
(0.18) (0.08) (0.05)
Wealth index: 3rd quintile 0.39 -0.13 -0.03
(0.19)** (0.07)* (0.06)
Wealth index: 4th quintile 0.06 -0.07 -0.01
(0.18) (0.07) (0.05)
Wealth index: 5th quintile 0.08 -0.08 -0.07
(0.17) (0.07) (0.05)
Baseline anemia measure 0.26 0.08 0.06
(0.06)*** (0.06) (0.07)
Constant 9.41 -0.06 0.03
(0.76)*** (0.14) (0.08)
R2 0.15 0.10 0.08
N 432 432 432
Notes: Standard errors in parenthesis clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. All estimations
contain age in years fixed effects and urban center fixed effects.
64
Table 8.12 Impact of treatment modalities on anemia measures among adolescent girls aged 10–16
covariates
Hemoglobin (g/dl) Any anemia Moderate or severe
Treatment==Food 0.05 0.06 -0.04
(0.14) (0.06) (0.04)
Treatment==Cash -0.09 0.02 -0.02
(0.13) (0.05) (0.04)
Treatment==Voucher -0.00 0.04 -0.04
(0.15) (0.05) (0.04)
R2 0.15 0.10 0.08
N 432 432 432
F test: Food=Voucher 0.12 0.09 0.03
P-value 0.73 0.77 0.87
F test: Cash=Voucher 0.38 0.13 0.32
P-value 0.54 0.72 0.57
F test: Food=Cash 0.99 0.42 0.53
P-value 0.32 0.52 0.47
Notes: Standard errors in parenthesis clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. All estimations
control for age in year splines, indicator of receiving an iron supplementation in the last 6 months, indicator of
currently menstruating, ethnicity, sex and education of household head, household size, number of children, wealth
quintiles and urban center fixed effects.
Table 8.13 Differential impact on anemia measures among adolescent girls aged 10–16 with covariates
with respect to ethnicity, by treatment arms
Hemoglobin (g/dl) Any anemia Moderate or severe
Treatment==Food 0.00 0.07 -0.03
(0.19) (0.07) (0.05)
Treatment==Cash -0.10 0.02 -0.02
(0.17) (0.06) (0.05)
Treatment==Voucher 0.04 0.05 -0.07
(0.18) (0.07) (0.05)
Food treatment X Colombian 0.22 -0.04 -0.06
(0.33) (0.14) (0.07)
Cash treatment X Colombian 0.01 -0.01 0.01
(0.30) (0.10) (0.08)
Voucher treatment X Colombian -0.13 -0.05 0.10
(0.35) (0.12) (0.09)
Household head is Colombian 0.25 -0.08 -0.10
(0.22) (0.08) (0.07)
R2 0.15 0.10 0.09
N 432 432 432
Notes: Standard errors in parenthesis clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. All estimations
control for age in year splines, indicator of receiving an iron supplementation in the last 6 months, indicator of
currently menstruating, ethnicity, sex and education of household head, household size, number of children, wealth
quintiles and urban center fixed effects.
65
9. Gender Issues
Women’s empowerment has been identified for decades as a development goal in its
own right, in addition to a means to achieve other targets, including poverty reduction (King
and Mason 2001). Accordingly, numerous cash and other transfer programs are designed with
the specific objective of empowering or placing resources in the hands of women in order to
have a greater impact on positive outcomes, particularly for children. Despite this
understanding, we still know relatively little about how transfer programs interact with
women’s empowerment within the household and how this dynamic may vary between
transfer types, cultures, and household structures. This evaluation provides an opportunity to
explore whether transfers affect women’s empowerment indicators including intrahousehold
decisionmaking power, disagreements, and intimate partner violence, and whether the impact
differs by modality.
We divide this chapter into two parts: we first discuss the analysis related to a woman’s
household decision-making and we then discuss the analysis related to intimate partner
violence. One woman per household over the age of 15 was administered the decision-making
and intimate partner violence questions, with priority given to “primary females” including
female heads of household, spouses of household heads, or next oldest female. While the
decision-making variables could be administered to women regardless of their relationship
status, the intimate partner violence questions were administered only if a woman had been in a
relationship in the past 6 months. Moreover, given the sensitive nature of these sections,
enumerators were instructed to administer the questions only if the woman was able to be
interviewed alone. In total, of the 2,122 panel households, we have a matched panel of 1,911
women who were administered the decisionmaking module at both the baseline and follow-up.
However, due to the restrictions on partnership, only 1,492 woman were administered the
domestic violence questions at baseline and 1,438 at follow-up. Of these women, we have a
matched panel for 1,268 who were administered the questions both at baseline and follow-up.
9.1 Decisionmaking
9.1.1 Decisionmaking Indicators
Although empowerment can be defined in a number of ways across different
disciplines, conceptualization generally refers to “women’s ability to make decisions and affect
outcomes of importance to themselves and their families (Malhotra, Schuler, and Boender
2002, 5).” Within this definition, researchers have focused on both direct and indirect measures
of empowerment. Direct measures generally focus on the expansion over time of a woman’s set
of available choices and the ability to transition these choices into desirable outcomes. Indirect,
or proxy, measures generally focus on the possession of resources, both tangible such as assets,
or intangible, such as education or social capital, which may then lead to or facilitate
empowerment. Although there are numerous methods of assessing whether or not the transfer
program had an impact on women’s empowerment, here we focus specifically on women’s
decisionmaking within the household. Women’s household decisionmaking across economic
and social spheres is often used in economic and public health literature to measure women’s
status, bargaining power, or empowerment within the household.
66
To measure women’s decisionmaking, we follow the approach used by the DHS, which
asks women to consider their relative decision-making power across a number of domains. In
both baseline and follow-up surveys, we ask who in the household generally has the final say in
decisions regarding (1) whether or not the woman works for pay, (2) children’s education,
(3) children’s health, (4) woman’s own health, (5) small daily food purchases, (6) large food
purchases, (7) large asset purchases (such as furniture, TV, etc.), (8) whether or not to use
contraceptives. The responses to these questions could be the following: (a) the woman herself,
(b) her spouse or partner, (c) the woman and spouse/partner together, (d) someone else in the
household, (e) the woman and someone else together, (d) the decision is not applicable (for
example, questions (2) and (3) in a household without children present). Given the not
applicable response option, the samples for each decisionmaking indicator vary based on the
response rates per question (ranging from 1,761 women answering questions regarding daily
food purchases to 1,238 women answering questions regarding children’s education).
In this analysis we focus on two main outcomes. First we present analysis of joint or sole
decisionmaking across domains, measuring the proportion of women with “any
decisionmaking power” in the domain. This measure is preferred to an index which ranks level
of involvement, as it is not clear from this perspective whether joint or independent
decisionmaking is preferred over the other. In addition to this main set of decisionmaking
questions, a number of other indicators were collected to corroborate and conduct robustness
checks on results. Our second indicator comes from this set of questions and indicates whether
or not there has been a disagreement over the decision domains listed above in the last six
months. This indicator is simply a yes or no answer and indicates if there have been changes in
opinions or challenges in intrahousehold decisionmaking in the recent past.
Table 9.1 shows the percentage of women exhibiting any decisionmaking power, and
experience of disagreements by sphere of decisionmaking at baseline. Focusing on the baseline
results, women report having highest involvement in decisionmaking over their own health
(92percent), children’s health (91 percent), children’s education and small daily food purchases
(both 87 percent). Women report relatively lower decisionmaking involvement in decisions
regarding their own work for pay (78 percent) and large purchase of assets (77 percent).
However it should be noted that these percentages over all decisionmaking domains are quite
high. The experience of disagreements across decisionmaking domains ranges from a high of 11
percent for own work, to lows of 3 percent for large food and for asset purchases. None of these
indicators show significant differences at the baseline between treatment and comparison
groups.
67
Table 9.1 Baseline means of decisionmaking indicators, by pooled treatment
All Comparison Treatment Difference
Sole or joint decisionmaking, own work for pay 0.78 0.77 0.79 -0.02
(0.01) (0.02) (0.01) ( 0.02)
Any disagreement, own work for pay 0.11 0.11 0.11 -0.00
(0.01) (0.01) (0.01) ( 0.02)
N 1,749 474 1,275
Sole or joint decisionmaking, children's education 0.87 0.88 0.87 -0.02
(0.01) (0.02) (0.01) ( 0.02)
Any disagreement, children's education 0.05 0.06 0.05 -0.00
(0.01) (0.01) (0.01) ( 0.02)
N 1,238 335 903
Sole or joint decisionmaking, children's health 0.91 0.90 0.91 -0.02
(0.01) (0.02) (0.01) ( 0.02)
Any disagreement, children's health 0.04 0.04 0.04 -0.00
(0.01) (0.01) (0.01) ( 0.02)
N 1,338 362 976
Sole or joint decisionmaking, own health 0.92 0.93 0.91 -0.02
(0.01) (0.01) (0.01) ( 0.02)
Any disagreement, own health 0.04 0.03 0.04 -0.00
(0.00) (0.01) (0.01) ( 0.02)
N 1,780 482 1,298
Sole or joint decisionmaking, daily food purchases 0.87 0.89 0.87 -0.02
(0.01) (0.01) (0.01) ( 0.02)
Any disagreement, daily food purchases 0.04 0.05 0.04 -0.00
(0.00) (0.01) (0.01) ( 0.02)
N 1,761 473 1,288
Sole or joint decisionmaking, large food purchases 0.80 0.81 0.80 -0.02
(0.01) (0.02) (0.01) ( 0.02)
Any disagreement, large food purchases 0.03 0.04 0.03 -0.00
(0.00) (0.01) (0.00) ( 0.02)
N 1,559 434 1,125
Sole or joint decisionmaking, purchase of large household assets 0.77 0.77 0.77 -0.02
(0.01) (0.02) (0.01) ( 0.02)
Any disagreement, purchase of large household assets 0.03 0.02 0.03 -0.00
(0.00) (0.01) (0.00) ( 0.02)
N 1,547 411 1,136
Notes: Sample sizes reflect number of women to which questions are applicable with responses over the panel period
Standard errors in parentheses. * p < 0.1 ** p < 0.05; *** p < 0.01.
Table 9.2 shows the percentage of women exhibiting any decision-making power, and
experience of disagreements by sphere of decisionmaking at follow-up. Overall, mean values
have not changed significantly during the program implementation period; however there are a
few exceptions. The percentage of women reporting disagreements with decisions to work for
pay show decreases (from 11 to 8 percent) in the treatment group and this difference is
statistically significant at the 5 percent level. In addition, the number of disagreements on child
health decisions decreases (from 4 to 2 percent) in the comparison group, and the percentage of
women involved in decisions regarding their own health decreases in the comparison group
(from 93 to 88 percent). However differences across comparison and treatment group are only
marginally significant at the 10 percent level.
68
Table 9.2 Follow-up means of decisionmaking indicators, by pooled treatment
All Comparison Treatment Difference
Sole or joint decisionmaking, own work for pay 0.79 0.78 0.79 -0.01
(0.01) (0.02) (0.01) ( 0.02)
Any disagreement, own work for pay 0.09 0.12 0.08 0.03
(0.01) (0.01) (0.01) ( 0.02)**
N 1,749 474 1,275
Sole or joint decisionmaking, children's education 0.85 0.85 0.85 -0.01
(0.01) (0.02) (0.01) ( 0.02)
Any disagreement, children's education 0.04 0.05 0.04 0.03
(0.01) (0.01) (0.01) ( 0.02)
N 1,238 335 903
Sole or joint decisionmaking, children's health 0.88 0.88 0.88 -0.01
(0.01) (0.02) (0.01) ( 0.02)
Any disagreement, children's health 0.04 0.02 0.04 0.03
(0.01) (0.01) (0.01) ( 0.02)*
N 1,338 362 976
Sole or joint decisionmaking, own health 0.90 0.88 0.91 -0.01
(0.01) (0.01) (0.01) ( 0.02)*
Any disagreement, own health 0.04 0.05 0.04 0.03
(0.00) (0.01) (0.01) ( 0.02)
N 1,780 482 1,298
Sole or joint decisionmaking, daily food purchases 0.88 0.87 0.88 -0.01
(0.01) (0.02) (0.01) ( 0.02)
Any disagreement, daily food purchases 0.04 0.04 0.04 0.03
(0.00) (0.01) (0.01) ( 0.02)
N 1,761 473 1,288
Sole or joint decisionmaking, large food purchases 0.80 0.78 0.81 -0.01
(0.01) (0.02) (0.01) ( 0.02)
Any disagreement, large food purchases 0.03 0.04 0.03 0.03
(0.00) (0.01) (0.00) ( 0.02)
N 1,559 434 1,125
Sole or joint decisionmaking, purchase of large household assets 0.76 0.73 0.77 -0.01
(0.01) (0.02) (0.01) ( 0.02)
Any disagreement, purchase of large household assets 0.02 0.03 0.02 0.03
(0.00) (0.01) (0.00) ( 0.02)
N 1,547 411 1,136
Notes: Sample sizes reflect number of women to which questions are applicable with responses over the panel period
Standard errors in parentheses. * p < 0.1 ** p < 0.05; *** p < 0.01.
9.1.2 Impact of Treatment on Decisionmaking Indicators
In Tables 9.3 and 9.4 we combine all three treatment arms (food, cash, and voucher) and
estimate the impact of pooled treatment on the indicator of any decisionmaking involvement
(sole or joint). We present the ANCOVA estimates with and without controlling for additional
covariates. We use the same modeling as in previous analysis; however we include additional
women-specific control variables, age, education, marital status, and ethnicity and exclude
some of the household-head specific control variables. As expected, given the relative success of
randomization, adding covariates has a marginal impact on the size of the coefficient on
treatment. As predicted from the descriptive analysis, the pooled treatment has no impact on
69
women’s decisionmaking involvement across all measures, with and without controlling for
other characteristics of the woman and household (Tables 9.3 and 9.4).
Similar to the decisionmaking analysis, Tables 9.5 and 9.6 present the combined effect of
the three treatment arms (food, cash, and voucher) on the indicators of any disagreements in the
past 6 months across decisionmaking domains with and without controlling for covariates. The
pooled treatment has a weakly significant and negative impact on disagreements regarding
decisionmaking for own work, and a weakly significant positive impact on disagreements
regarding decisionmaking on children’s health. However, when controlling for covariates,
pooled treatment no longer has any significant relationship with experience of disagreements of
any type.
Tables 9.7 and 9.8 replicate the ANCOVA estimates for decisionmaking and
disagreements, for each treatment arm separately, and conduct Wald tests to examine whether
the estimates from each treatment arm are significantly different from each other. Specifications
include a full set of control variables; however, for simplicity we only present the coefficients
from the different treatment arms. Table 9.7 shows that none of the treatment arms have
significant impacts on decisionmaking, while Table 9.8 shows the only significant impact is a
positive impact by the food arm on experience of disagreements regarding child health.
In conclusion, overall we find little evidence of program impact on women’s
decisionmaking as measured by involvement in any decisions across a number of social and
economic domains. We also find little evidence of program impact on the alternative measure of
women’s experience of disagreements across decisionmaking domains. These results do not
vary when we consider interactions with Colombian ethnicity (and thus omit them here). There
are number of potential explanations for this lack of significant effect. First, although
decisionmaking has often been operationalized in a similar way as we have done, these
measures may not be sensitive or specific enough to identify an effect. In addition, since the
program is a relatively short time period, it may not be enough to change longer-term
decisionmaking dynamics within a household.
Table 9.3 Impact of pooled treatment on sole or joint decisionmaking
Own
work
Children's
education
Children's
health
Own
health
Small food
purchases
Large food
purchases
Asset
purchases
Pooled treatment 0.01 -0.00 0.00 0.03 0.02 0.03 0.04
(0.02) (0.03) (0.02) (0.02) (0.02) (0.03) (0.03)
Baseline measure 0.24 0.18 0.17 0.18 0.19 0.21 0.19
(0.03)*** (0.04)*** (0.04)*** (0.04)*** (0.03)*** (0.03)*** (0.03)***
Constant 0.60 0.70 0.72 0.72 0.70 0.61 0.59
(0.03)*** (0.05)*** (0.05)*** (0.05)*** (0.04)*** (0.04)*** (0.03)***
R2 0.06 0.03 0.02 0.03 0.04 0.05 0.04
N 1,749 1,238 1,338 1,780 1,761 1,559 1,547
Notes: Standard errors in parenthesis clustered at the cluster level. * p < 0.1; ** p < 0.05; *** p < 0.01.
70
Table 9.4 Impact of pooled treatment on sole or joint decisionmaking, with covariates
Own
work
Children's
education
Children's
health
Own
health
Small food
purchases
Large food
purchases
Asset
purchases
Pooled treatment -0.01 -0.01 0.01 0.02 -0.02 0.00 0.02
(0.02) (0.03) (0.02) (0.02) (0.02) (0.03) (0.03)
Colombian -0.03 -0.00 0.00 -0.02 -0.01 -0.02 0.01
(0.02) (0.02) (0.02) (0.02) (0.02) (0.03) (0.02)
Age 0.00 -0.00 -0.00 -0.00 -0.00 0.00 -0.00
(0.00) (0.00) (0.00)** (0.00) (0.00) (0.00) (0.00)
Secondary education or higher 0.02 -0.00 -0.03 -0.01 -0.01 0.00 0.02
(0.02) (0.02) (0.02) (0.01) (0.02) (0.02) (0.02)
Marital status: Married -0.05 -0.05 -0.00 0.00 0.07 0.03 -0.00
(0.03)* (0.03)* (0.02) (0.02) (0.02)*** (0.03) (0.03)
Marital status: Single 0.18 0.10 0.10 0.09 0.13 0.20 0.22
(0.02)*** (0.03)*** (0.02)*** (0.02)*** (0.02)*** (0.02)*** (0.03)***
Marital status: Divorced or separated 0.17 0.08 0.06 0.07 0.10 0.17 0.21
(0.02)*** (0.03)** (0.03)** (0.02)*** (0.02)*** (0.03)*** (0.03)***
Marital status: Widowed 0.14 0.02 0.05 0.05 0.08 0.11 0.11
(0.04)*** (0.05) (0.04) (0.03)* (0.03)** (0.04)*** (0.05)**
Indigenous 0.02 -0.01 0.02 -0.02 -0.01 0.02 0.00
(0.05) (0.05) (0.04) (0.04) (0.04) (0.05) (0.06)
Afro-Ecuadorian 0.01 -0.04 -0.09 0.01 0.02 0.05 0.03
(0.03) (0.04) (0.04)** (0.03) (0.03) (0.03) (0.04)
Household size -0.01 -0.02 -0.02 -0.01 -0.00 -0.01 -0.02
(0.01) (0.01) (0.01) (0.01)* (0.01) (0.01) (0.01)
Number of children 0-5 years 0.01 -0.00 -0.01 0.01 -0.02 0.00 -0.01
(0.02) (0.02) (0.02) (0.01) (0.02) (0.02) (0.02)
Number of children 6-15 years 0.02 0.01 0.02 0.01 0.01 0.01 0.02
(0.01) (0.02) (0.01) (0.01) (0.01) (0.01) (0.02)
Wealth index: 2nd quintile -0.05 -0.03 -0.03 -0.02 0.01 0.02 0.01
(0.03) (0.04) (0.03) (0.02) (0.03) (0.04) (0.03)
Wealth index: 3rd quintile -0.03 -0.02 -0.01 0.01 -0.00 -0.00 -0.01
(0.03) (0.04) (0.03) (0.02) (0.03) (0.04) (0.03)
Wealth index: 4th quintile -0.05 -0.02 -0.01 -0.02 -0.02 -0.03 -0.05
(0.03) (0.04) (0.03) (0.03) (0.03) (0.04) (0.03)
Wealth index: 5th quintile -0.04 0.00 0.00 -0.01 -0.04 -0.01 -0.00
-0.01 -0.01 0.01 0.02 -0.02 0.00 0.02
Baseline measure 0.17 0.14 0.14 0.15 0.16 0.16 0.13
(0.03)*** (0.04)*** (0.05)*** (0.04)*** (0.03)*** (0.03)*** (0.03)***
Constant 0.63 0.90 0.93 0.86 0.78 0.66 0.71
(0.06)*** (0.08)*** (0.08)*** (0.05)*** (0.06)*** (0.07)*** (0.07)***
R2 0.11 0.07 0.08 0.06 0.07 0.09 0.11
N 1,749 1,238 1,338 1,780 1,761 1,559 1,547
Notes: Standard errors in parenthesis clustered at the cluster level. * p < 0.1; ** p < 0.05; *** p < 0.01. All estimations
contain urban center fixed effects. Omitted categories are: Ecuadorian or mixed, in consensual union, primary or no
education. Wealth index constructed through principal component analysis using asset ownership and dwelling
characteristics.
71
Table 9.5 Impact of pooled treatment on disagreements regarding decisionmaking
Own
work
Children's
education
Children's
health
Own
health
Small food
purchases
Large food
purchases
Asset
purchases
Pooled Treatment -0.03 -0.01 0.02 -0.01 -0.00 -0.01 -0.00
(0.02)* (0.02) (0.01)* (0.01) (0.01) (0.01) (0.01)
Baseline measure 0.11 0.07 0.02 0.07 0.05 0.01 0.09
(0.03)*** (0.04)* (0.03) (0.04)* (0.03) (0.03) (0.05)**
Constant 0.11 0.04 0.02 0.05 0.04 0.04 0.02
(0.02)*** (0.01)*** (0.01)** (0.01)*** (0.01)*** (0.01)*** (0.01)***
R2 0.02 0.01 0.00 0.01 0.00 0.00 0.01
N 1,749 1,238 1,338 1,780 1,761 1,559 1,547
Notes: Standard errors in parenthesis clustered at the cluster level. * p < 0.1; ** p < 0.05; *** p < 0.01.
Table 9.6 Impact of pooled treatment on disagreements regarding decisionmaking, with covariates
Own
work
Children's
education
Children's
health
Own
health
Small food
purchases
Large food
purchases
Asset
purchases
Pooled treatment -0.02 -0.00 0.02 0.00 0.00 -0.01 0.00
(0.02) (0.02) (0.01) (0.01) (0.01) (0.01) (0.01)
Colombian 0.00 -0.01 -0.01 0.00 0.02 -0.00 -0.02
(0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01)**
Age -0.00 0.00 -0.00 -0.00 -0.00 0.00 -0.00
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Secondary education or higher 0.01 0.01 0.01 0.01 -0.01 0.01 -0.01
(0.02) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01)
Marital status: Married -0.03 0.01 0.03 -0.01 -0.01 -0.01 0.01
(0.02) (0.01) (0.01)* (0.01) (0.01) (0.01) (0.01)
Marital status: Single -0.09 -0.05 -0.03 -0.05 -0.05 -0.04 -0.03
(0.02)*** (0.01)*** (0.01)*** (0.01)*** (0.01)*** (0.01)*** (0.01)***
Marital status: Divorced or separated -0.09 -0.03 -0.02 -0.04 -0.04 -0.04 -0.03
(0.02)*** (0.02)* (0.01)* (0.01)** (0.01)** (0.01)*** (0.01)***
Marital status: Widowed -0.06 -0.02 -0.02 -0.03 -0.04 -0.04 -0.02
(0.03)* (0.03) (0.01)* (0.02)* (0.02)* (0.01)*** (0.01)*
Indigenous -0.04 0.03 -0.01 -0.01 -0.04 -0.01 -0.02
(0.03) (0.03) (0.02) (0.02) (0.02)** (0.02) (0.01)***
Afro-Ecuadorian -0.02 0.02 0.03 -0.02 -0.02 -0.00 0.01
(0.02) (0.04) (0.03) (0.01)* (0.01) (0.02) (0.02)
Household size -0.00 -0.01 0.00 0.00 -0.01 -0.00 -0.00
(0.01) (0.01) (0.00) (0.00) (0.00)* (0.00) (0.00)
Number of children 0-5 yrs 0.01 0.01 0.00 0.00 0.01 0.01 0.01
(0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01)*
Number of children 6-15 yrs 0.01 0.02 0.00 0.00 0.01 0.01 0.00
(0.01) (0.01)** (0.01) (0.01) (0.01)** (0.01) (0.00)
Wealth index: 2nd quintile 0.03 0.01 -0.00 -0.01 -0.00 0.01 0.00
(0.02) (0.02) (0.02) (0.01) (0.01) (0.01) (0.01)
Wealth index: 3rd quintile 0.04 -0.01 0.02 0.00 0.02 0.02 0.02
(0.02)* (0.01) (0.02) (0.01) (0.01) (0.01) (0.01)*
Wealth index: 4th quintile 0.03 -0.02 -0.04 -0.01 -0.00 0.01 0.01
(0.02) (0.02) (0.01)*** (0.01) (0.01) (0.01) (0.01)
Wealth index: 5th quintile 0.04 -0.00 0.00 0.03 0.02 0.03 0.02
(0.02)* (0.02) (0.02) (0.02) (0.02) (0.01)** (0.01)
Baseline measure 0.08 0.06 0.01 0.06 0.04 0.00 0.08
(0.03)*** (0.04) (0.03) (0.04) (0.03) (0.03) (0.05)*
Constant 0.12 0.03 0.02 0.03 0.06 0.04 0.04
72
(0.05)** (0.03) (0.03) (0.03) (0.03)** (0.02)* (0.02)**
R2 0.04 0.02 0.03 0.03 0.02 0.02 0.03
N 1,749 1,238 1,338 1,780 1,761 1,559 1,547
Notes: Standard errors in parenthesis clustered at the cluster level. * p < 0.1; ** p < 0.05; *** p < 0.01. All estimations
contain urban center fixed effects. Omitted categories are: Ecuadorian or mixed, in consensual union, primary or no
education. Wealth index constructed through principal component analysis using asset ownership and dwelling
characteristics.
Table 9.7 Impact of treatment modalities on sole or joint decisionmaking
Own
work
Children's
education
Children's
health
Own
health
Small food
purchases
Large food
purchases
Asset
purchases
Treatment==Food -0.01 0.03 0.03 0.01 -0.02 0.01 0.03
(0.03) (0.03) (0.03) (0.02) (0.03) (0.03) (0.03)
Treatment==Cash -0.01 -0.03 -0.01 0.02 -0.00 -0.01 -0.01
(0.03) (0.04) (0.03) (0.02) (0.03) (0.03) (0.03)
Treatment==Voucher -0.00 -0.01 0.01 0.02 -0.02 0.01 0.03
(0.03) (0.04) (0.03) (0.02) (0.02) (0.03) (0.03)
R2 0.11 0.07 0.08 0.06 0.07 0.09 0.11
N 1,749 1,238 1,338 1,780 1,761 1,559 1,547
F test: Food=Voucher 0.05 1.58 0.55 0.25 0.00 0.00 0.02
P-value 0.83 0.21 0.46 0.61 0.98 0.97 0.90
F test: Cash=Voucher 0.06 0.18 0.64 0.05 0.58 0.56 1.36
P-value 0.81 0.67 0.43 0.83 0.45 0.45 0.25
F test: Food=Cash 0.00 3.11 2.39 0.06 0.63 0.48 2.39
P-value 0.98 0.08 0.12 0.80 0.43 0.49 0.12
Note: Standard errors in parenthesis clustered at the cluster level. * p < 0.1; ** p < 0.05; *** p < 0.01. All estimations
contain baseline outcome variable, urban center fixed effects, and a full set of control variables.
Table 9.8 Impact of treatment modalities on disagreements regarding decisionmaking domains
Own
work
Children's
education
Children's
health
Own
health
Small food
purchases
Large food
purchases
Asset
purchases
Treatment==Food -0.03 -0.01 0.03 0.00 0.02 0.01 0.00
(0.03) (0.02) (0.01)** (0.02) (0.02) (0.02) (0.01)
Treatment==Cash -0.01 -0.01 0.01 0.01 0.00 -0.01 0.00
(0.02) (0.02) (0.01) (0.02) (0.02) (0.01) (0.01)
Treatment==Voucher -0.03 0.00 0.02 -0.01 -0.00 -0.01 0.01
(0.02) (0.02) (0.02) (0.01) (0.01) (0.01) (0.01)
R2 0.04 0.03 0.03 0.03 0.02 0.02 0.03
N 1,749 1,238 1,338 1,780 1,761 1,559 1,547
F test: Food=Voucher 0.04 0.81 0.15 0.71 2.37 3.18 0.13
P-value 0.84 0.37 0.70 0.40 0.13 0.08 0.72
F test: Cash=Voucher 0.76 0.52 1.16 3.32 0.50 0.06 0.09
P-value 0.38 0.47 0.28 0.07 0.48 0.81 0.77
F test: Food=Cash 0.27 0.03 2.71 0.42 0.80 2.45 0.01
P-value 0.60 0.86 0.10 0.52 0.37 0.12 0.92
Notes: Standard errors in parenthesis clustered at the cluster level. * p < 0.1; ** p < 0.05; *** p < 0.01. All estimations
contain baseline outcome variable, urban center fixed effects and a full set of control variables.
73
9.2 Intimate Partner Violence
9.2.1 Intimate Partner Violence Indicators
Intimate partner violence is a multidimensional and complex issue that is usually
categorized into physical violence, psychological violence, and sexual violence. We use
questions from WHO’s Violence Against Woman Instrument to assess intimate partner violence
in our sample population (Garcia-Moreno et al, 2005). This instrument has been tested and
validated by WHO in multiple settings and countries. The questions on partner violence
explore aspects of controlling behaviours, emotional abuse, physical violence, and sexual
violence. The instrument does not aim to document every abusive action that a woman may
experience, but rather aims to maximize disclosure. We follow WHO and DHS protocol and
construct indicators for controlling behaviors, emotional violence, and physical violence and/or
sexual violence.
The questionnaire includes seven questions on physical violence: whether a woman has
been (1) pushed, shoved, or had an object thrown; (2) slapped or had her arm twisted; (3)
punched or hit with object; (4) kicked or dragged; (5) strangled or burned; (6) attacked with
knife or other weapon; and (7) threatened to be attacked with knife or other weapon. For sexual
violence, there are two questions that are asked at both baseline and follow-up: whether woman
has been (1) forced to have sexual interactions, and (2) forced to conduct sexual acts that woman
does not approve. We create a physical and/or sexual violence indicator that equals 1 if a
woman reports yes to having experienced any type of physical or sexual violence in the last
6 months (approximately the length of the intervention period).
In the baseline survey there are three questions that can be categorized as “emotional
violence”: whether a woman has been (1) humiliated or insulted, (2) threatened to be
abandoned, and (3) threatened to have children taken away. In addition to these three
questions, the follow-up survey added the following two indicators: (4) whether partner
threatened to hurt woman or someone she knows, and (5) whether woman has been humiliated
or insulted in front of others. We use all 5 questions administered at follow-up to create an
indicator for emotional violence that equals 1 if a woman reports yes to having experienced any
of the five emotional violence indicators in the last 6 months. For controlling behaviors there are
two questions at baseline and three additional questions that were added at follow-up: whether
a partner (1) accuses woman of being unfaithful, (2) limits woman’s contact with her family, (3)
limits woman’s contact with friends, (4) wants to know where the woman is at all times, and (5)
ignores or is indifferent to woman. Similar to the emotional violence indicator, we use all five
questions that were administered at follow-up to create an indicator for controlling behaviors
that equals 1 if a woman reports yes to having experienced any of the five controlling behaviors
indicators in the last 6 months.
We also combine all types of intimate partner violence and create an indicator that
equals one if a woman experiences any controlling behaviors, emotional violence, physical
violence, and sexual violence. We define this indicator as “Any violence” in the analysis.
At baseline, 17 percent of woman experience controlling behaviors by partner in the last
6 months, 26 percent experience emotional violence, 16 percent experience physical and/or
sexual violence, and 33 percent experience any type of violence in the past 6 months (Table 9.9).
There are no significant differences in means across treatment and comparison arms for
74
controlling behaviors, or emotional violence; however, the treatment arm experienced
significantly higher physical and/or sexual violence.
The percent of women experiencing controlling behaviors or emotional violence at
baseline is lower than at follow-up and this most likely reflects the fact that at baseline there
were fewer questions administered (Table 9.10). Although the percent of women experiencing
controlling behaviors increases for both the comparison and treatment arm, the increase is
much larger for the comparison arm. Consequently, the difference in means between the
comparison and treatment arm for controlling behaviors is now statistically significant. The
comparison arm also experiences significantly higher rates of physical and sexual violence and
any violence.
The large rise in intimate partner violence amongst the comparison group could be due
to either an upward secular trend in violence or to resentment of partners taken out on woman
for not receiving the transfer. Closer inspection of the data reveals that this large increase in
violence from the comparison arm is mainly due to an extraordinary increase in violence in one
urban center. While we conduct our analysis with and without this urban center, we intend to
further explore the issue of a rise in violence in the comparison group with a qualitative study
in the Fall 2012.
Table 9.9 Intimate partner violence baseline means, by pooled treatment
All Comparison Treatment Difference
Controlling 0.17 0.17 0.17 -0.00
(0.01) (0.02) (0.01) ( 0.02)
Emotional 0.26 0.24 0.26 0.02
(0.01) (0.02) (0.01) ( 0.03)
Physical and or sexual 0.16 0.13 0.17 0.05
(0.01) (0.02) (0.01) ( 0.02)**
Any 0.33 0.31 0.33 0.03
(0.01) (0.02) (0.02) ( 0.03)
N 1,268 357 911
Notes: Standard errors in parentheses. * p < 0.1 ** p < 0.05; *** p < 0.01.
Table 9.10 Intimate partner violence follow-up means, by pooled treatment
All Comparison Treatment Difference
Controlling 0.35 0.41 0.32 -0.09
(0.01) (0.03) (0.02) ( 0.03)***
Emotional 0.28 0.30 0.28 -0.02
(0.01) (0.02) (0.01) ( 0.03)
Physical and or sexual 0.17 0.20 0.15 -0.05
(0.01) (0.02) (0.01) ( 0.02)*
Any 0.41 0.46 0.39 -0.07
(0.01) (0.03) (0.02) ( 0.03)**
N 1,268 357 911
Notes: Standard errors in parentheses. * p < 0.1 ** p < 0.05; *** p < 0.01.
75
9.2.2 Impact of Treatment on Intimate Partner Violence Indicators
Tables 9.11 and 9.12 present the ANCOVA estimates of being in the treatment arm on
intimate partner violence indicators. In Table 9.11, we combine all three treatment arms (food,
cash, and voucher) and estimate the impact of pooled treatment on controlling behaviors,
emotional violence, physical and/or sexual violence, and any violence. We present the results
with and without controlling for any covariates. Given that the coefficients change slightly
when we add covariates, we focus on columns 2, 4, 6, and 8 of Table 9.11. Pooled treatment
significantly decreases controlling behaviors, physical/sexual violence, and any violence. In
particular, pooled treatment leads to an 8 percentage point decrease in controlling behaviors,
and a 7 percentage point decrease in physical/sexual violence and any violence. However, as
tables 9.10 and 9.11 reveal, most of this decrease is due to large increases in violence from the
comparison arm from baseline to follow-up. Removing from the analysis the urban center with
the strong upward trend in the comparison group, produces similar results, however, the
coefficient on treatment for physical/sexual violence and any violence decreases to 3 percentage
points. Other factors that contribute significantly to any violence are education, being
Colombian, married, age, and wealth.
Table 9.12 presents the impact estimates for each treatment arm separately, and
conducts Wald tests to exam whether the estimates from each treatment arm are significantly
different from each other. Similar to tables 9.7 and 9.8 we report only treatment coefficients,
although all estimates control for the full set of covariates. All three modalities (food, cash, and
voucher) significantly decrease physical/sexual violence. Only food and cash significantly
decrease controlling behaviors although the size of the coefficient is similar across modalities.
Only cash significantly decreases any violence, although again the size of the coefficient is
similar across modalities. Removing the urban center with the upward trend in violence among
the comparison group again produces similar results, but the coefficients for physical/sexual
violence decrease to 6 percentage points for the food arm, 1 percentage points for the cash arm,
and 2 percentage points for the voucher arm.
76
Table 9.11 Impact of pooled treatment on intimate partner violence measures
Controlling Emotional Physical/Sexual Any
(1) (2) (3) (4) (5) (6) (7) (8)
Pooled Treatment -0.09 -0.08 -0.03 -0.01 -0.06 -0.07 -0.08 -0.07
(0.04)** (0.04)** (0.03) (0.03) (0.03)** (0.03)** (0.04)** (0.04)*
Colombian -0.05 -0.05 -0.03 -0.05
(0.03) (0.03) (0.02) (0.03)*
Age -0.00 0.00 -0.00 -0.00
(0.00)** (0.00) (0.00) (0.00)*
Secondary education or higher -0.06 -0.06 0.01 -0.06
(0.03)** (0.03)** (0.02) (0.03)**
Married -0.06 -0.08 -0.05 -0.07
(0.03)* (0.03)*** (0.02)** (0.03)**
Indigenous 0.03 0.02 -0.01 0.07
(0.08) (0.07) (0.05) (0.08)
Afro-Ecuadorian -0.03 -0.03 0.01 -0.02
(0.06) (0.05) (0.04) (0.06)
Household size 0.01 -0.01 -0.00 0.01
(0.01) (0.01) (0.01) (0.01)
Number of children 0-5 years 0.01 0.07 0.01 0.01
(0.02) (0.02)*** (0.02) (0.02)
Number of children 6-15 years -0.01 0.03 0.01 0.00
(0.02) (0.02)* (0.01) (0.02)
Wealth index: 2nd quintile 0.01 0.02 -0.03 -0.00
(0.04) (0.04) (0.04) (0.05)
Wealth index: 3rd quintile 0.06 0.05 -0.02 0.05
(0.04) (0.04) (0.03) (0.05)
Wealth index: 4th quintile 0.06 0.08 0.02 0.06
(0.04) (0.04)** (0.04) (0.05)
Wealth index: 5th quintile 0.09 0.06 -0.00 0.10
(0.05)* (0.04) (0.03) (0.05)**
Baseline controlling 0.35 0.35
(0.04)*** (0.04)***
Baseline emotional 0.36 0.35
(0.03)*** (0.03)***
Baseline physical and or sexual 0.37 0.37
(0.03)*** (0.03)***
Baseline any 0.33 0.33
(0.03)*** (0.03)***
Constant 0.35 0.34 0.21 0.10 0.15 0.19 0.36 0.35
(0.03)*** (0.08)*** (0.03)*** (0.07) (0.03)*** (0.07)*** (0.03)*** (0.08)***
R2 0.08 0.11 0.12 0.16 0.14 0.15 0.11 0.14
N 1,268 1,268 1,268 1,268 1,268 1,268 1,268 1,268
Notes: Standard errors in parentheses clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01. Columns 2, 4, 6,
and 8 contain urban center fixed effects. Wealth index constructed through principal component analysis using asset
ownership and dwelling characteristics.
77
Table 9.12 Impact of treatment modalities on intimate partner violence measures
Controlling Emotional Physical/sexual Any
Treatment==Food -0.07 -0.00 -0.08 -0.05
(0.04)* (0.04) (0.04)** (0.05)
Treatment==Cash -0.10 -0.03 -0.07 -0.09
(0.05)** (0.04) (0.04)* (0.05)*
Treatment==Voucher -0.06 -0.00 -0.07 -0.06
(0.04) (0.04) (0.04)* (0.05)
R2 0.12 0.16 0.15 0.14
N 1,268 1,268 1,268 1,268
F test: Food=Voucher 0.16 0.00 0.25 0.02
P-value 0.69 0.96 0.62 0.90
F test: Cash=Voucher 1.48 0.40 0.03 0.59
P-value 0.23 0.53 0.86 0.44
F test: Food=Cash 0.59 0.44 0.43 0.72
P-value 0.44 0.51 0.51 0.40
Notes: Standard errors in parenthesis clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01.
All estimations contain baseline outcome variable, urban center fixed effects, and a full set of
control variables.
Our results do not vary when we consider interactions with Colombian nationality (and thus
we omit them here). However, an interesting extension is to investigate whether the results
differ by whether a male or female receive the transfer. Given that we only know who normally
received the transfer for treatment households that took up the program, our estimates are no
longer intent-to-treat estimates, and thus our sample size decreases slightly. Table 9.13 reveals
that the impact on violence does not depend on the sex of the recipient.
Table 9.13 Impact of treatment by gender on intimate partner violence measures
Controlling Emotional Physical/sexual Any
Treatment, male recipient -0.10 -0.07 -0.11 -0.10
(0.05)** (0.04) (0.04)*** (0.05)**
Treatment, female recipient -0.09 -0.01 -0.07 -0.07
(0.04)** (0.03) (0.03)** (0.04)*
R2 0.12 0.17 0.16 0.15
N 1,063 1,063 1,063 1,063
F test: Male=Female 0.11 2.52 2.20 0.37
P-value 0.74 0.11 0.14 0.54
Notes: Standard errors in parenthesis clustered at the cluster level. * p < 0.1 ** p < 0.05; *** p < 0.01.
All estimations contain baseline outcome variable, urban center fixed effects, and a full set of
control variables.
78
10. Costing and Cost-Effectiveness
10.1 Costing Methods and Results
An important question to address in assessing the relative effectiveness of different food
assistance modalities is the cost of implementing each modality. A relative assessment of cost-
effectiveness by modality allows an examination of which mechanism (cash, voucher, or food)
provides the greatest benefit for the amount of funds invested. The particular goals of this
costing analysis are to answer the following two research questions: (1) What are the relative costs
of each modality (cash, voucher, and food)? (2) Which modalities are the most cost-effective?
While WFP tracks program costs via traditional accounting for its own records and for
external accountability purposes, such methods do not allow for an accurate breakdown by
modality. Traditional accounting costs often underestimate the true overall cost of program
operations due to, among other things, the cost of staff time dedicated to each treatment type.
Therefore, the Activity-based Costing – Ingredients (ABC-I) approach is used to calculate costs
for the analysis. The ABC-I method is a combination of activity-based accounting methods with
the “ingredients” method, which calculates program costs from inputs, input quantities, and
input unit costs (Fiedler, Villalobos, and de Mattos 2008; Tan-Torres Edejer et al. 2003). As the
ingredients method alone does not allocate costs according to program activities, it does not
allow for comparison between modalities. However, this method, when paired with the ABC
approach, matches activities with all their corresponding inputs into cost centers. The use of the
ABC-I method allows for opportunity costs, quantified as economic costs, to be included in the
total program costs. This method also allows for the incorporation of “off-budget”
expenditures, for example, donated goods or services that otherwise would not be included as
program operating costs.
The costing analysis utilizes data from the WFP-CO accounting ledger, information
gathered from staff by the CO finance department, internal procurement and operations
documents, as well as interviews with local partners. An advantage of the detailed information
on costs from the WFP accounting ledgers is that it permits the separation of costs that are
common across program modalities from those that are modality specific. A second strength of
the cost data is that we can calculate the staff costs associated with the intervention. Distinct
cost calculations are necessary to allow for inclusion of actual operational field costs, as well as
to avoid double-counting. For example, contracts with collaborating partners were developed as
part of the Protracted Relief and Recovery Operation (PRRO), an umbrella program that
provides the food component of the food, cash, and voucher program. However, the PRRO also
includes the provision of rations for a broader food intervention, also run by WFP-CO. In order
to assure that the cost excludes those activities that support the PRRO rather than just the
transfer program, a proportional amount is calculated of contracting cost in relation to the total
amount of metric tons of food for the transfer program in comparison with total tonnage for
PRRO. The complete calculations, including a detailed breakdown of cost by line item, are
included in Appendix 2 and Appendix 3. Final calculations are separated into two formats, one
representing total budgeted costs, and the other representing total actual costs.
WFP-CO staff provided cost calculations for five months of program activity. We
extrapolate the sixth month of costs from the previous five months’ data for those activities that
79
continued into the last month of the program. While the total number of transfers varied
slightly from month to month, the mean values for number of beneficiaries over the six-month
period are utilized for the cost-per-beneficiary analysis.
The most costly modality to implement was the food transfer (Figure 10.1). In particular,
it cost $14.36 to transfer a $40 voucher to a beneficiary, $14.77 to provide them with $40 in cash,
and $25.93 to transfer $40 worth of food. These numbers represent the total cost of
implementation minus the transfer cost. In total, over the six transfer periods, it cost $86.17 per
voucher beneficiary, $88.62 per cash beneficiary, and $155.64 per food beneficiary to provide
transfers. Over time, the cost of implementing these types of transfer programs may decline due
to economies of scale. However, that phenomenon will principally apply to cash and voucher
transfers, as several of the principal costs for the food transfer are costs that will not decrease as
the number of beneficiaries increase. For example, if one was to increase the number of voucher
or cash beneficiaries, the additional cost of producing extra debit cards or printing more
vouchers is minimal, and some costs may decrease over time. However, in the case of food, the
cost to add beneficiaries will not decrease, as costs like ration packaging are fixed sums per
ration distributed.
Figure 10.1 Total cost to transfer $40, by modality
Figure 10.2 depicts the cost breakdown, or disaggregation, for the per transfer cost by
modality. Costs are separated into two categories, modality specific, and non-modality specific,
in order to pinpoint which modality-related activities has the most impact on cost. Non-
modality specific costs are costs that are common across all modalities, such as materials for the
beneficiary trainings, or administrative costs for sub-offices. Primarily, the small differences in
non-modality specific costs between modalities originate from differences in the number of
beneficiaries, in that some cost categories had to be weighted to reflect the number of
beneficiaries for that modality type. The cash and voucher modalities have nearly identical
percentage cost breakdowns. However, the cost activities are different between the two. For
example, the cash transfer incurs modality-specific costs in the form of fees charged by the bank
to emit an ATM card for each beneficiary. The voucher modality-specific costs include voucher
design, liquidation and execution of payment. The food transfer has significantly higher
$0.00
$10.00
$20.00
$30.00
$40.00
Voucher Cash Food
$14 $15
$26
$21 $23
$36
Co
st in
USD
Actual Budgeted
80
modality-specific costs (44 percent of the total cost per transfer, in contrast with cash or
voucher, at 20 percent and 22 percent, respectively). The high cost of the food transfer is due to
storage costs, ration preparation, and ration distribution.
Figure 10.2 Modality-specific and nonspecific costs
Figure 10.3 shows the differences in cost types for human resources versus physical
resources. Human resources are categorized as staff time allocated to activities. Cash and food
require a similar percentage of human resource cost to physical cost which is less than that of
vouchers. The higher human resource cost of the voucher appears to originate from operational
activities conducted by WFP staff, such as voucher design. The voucher design, distribution,
liquidation and supermarket selection were all tasks directly handled by WFP, and the latter
two were recurring monthly costs. In the case of cash and food, much operations labor was
provided by partners.
If one were to “cash out,” or switch the least expensive modality (voucher, at $86.17 per
beneficiary) for the most expensive modality (food, at $155.64 per beneficiary), 739 extra
beneficiaries could be included in the program. If the food transfers were exchanged for cash,
693 beneficiaries could be added to the program. These costs are calculated from the actual cost
calculations. These calculations, however, are from a pure cost standpoint and do not
incorporate the impacts of each modality relative to cost.
$0.00
$5.00
$10.00
$15.00
$20.00
$25.00
$30.00
Voucher Cash Food
Non-specific $11.06 $11.74 $14.43
Modality specific $3.30 $3.03 $11.50
Co
st in
USD
81
Figure 10.3 Total cost by type (percent), by modality
10.2 Cost-Effectiveness
In order to compare the cost-effectiveness across modalities, we use the modality
specific costs from Figure 10.2 and the impact estimates for food security outcomes (Chapter 6).
We use modality specific costs because we are interested in the marginal cost of implementing
each modality type. In other words, after all common costs of program implementation
(planning, targeting, sensitization, nutrition trainings, etc.) are accounted for, what additional
costs are incurred to deliver these transfers in the form of food, cash, or voucher. Expressing
these in per transfer terms, it cost $11.50 to provide a food transfer, $3.03 to provide a cash
transfer, and $3.30 to provide a voucher transfer. We combine these costs with impact estimates
of the value of food consumption and caloric intake, as well as dietary diversity indices: the
HDDS, the DDI and the FCS (see Chapter 6 for definitions of these indices and impact
estimates).
In order to compare the cost-effectiveness across modalities, we do a simulation
whereby beneficiaries’ food security outcomes increase by 15 percent. Specifically, we calculate
how much it would cost to achieve this goal using food, cash, and vouchers, conditional on the
transfer size and abstracting from costs common to all modalities. Given the different metrics by
which our outcomes are measured, we conduct the simulations for each outcome separately.
For example, Figure 6.2 tells us that cash transfers increase the FCS by approximately 11
percent. Therefore, the modality specific cost of increasing FCS by 15 percent using cash
transfers is (15%/11%) X $3.03, which equals $4.13.
Table 10.1 shows the results of these calculations for each modality for the following five
outcomes: the value of per capita food consumption, per capita caloric intake, HDDS, DDI, and
FCS. There are two key findings. First, across all outcomes food is always the most costly means
of improving these outcomes by 15 percent. Second, vouchers are usually the least costly means
of improving these outcomes by 15 percent, although for increasing the value of food
consumption, there is virtually no difference in the cost of vouchers versus cash.
0 10 20 30 40 50 60 70 80 90 100
Vo
uch
er
Cas
hFo
od
Percentage cost
Mo
dal
ity
typ
e
Human Resources
Physical Resources
82
Table 10.1 Modality-specific cost of improving food security outcomes by 15 percent
Food Cash Voucher
Consumption $10.78 $3.79 $3.81
Calories $10.78 $7.58 $4.50
HDDS $28.75 $11.36 $8.25
DDI $15.68 $3.25 $2.91
FCS $17.25 $4.13 $3.09
Note: Modality-specific costs per transfer are used to calculate the cost of increasing each outcome by 15 percent.
83
11. Qualitative Lessons on Nutrition Training, Knowledge, and Behavior Change
Upon request by the WFP-CO, IFPRI developed a small-scale qualitative study designed
to examine the efficacy of nutrition trainings that accompanied the distribution of the different
transfer types. This study was proposed as a complement to the quantitative survey in order to
explore beneficiary preference, knowledge retention, and adoption of nutrition training
components. Key research questions included the following: What are beneficiaries learning
from nutritional trainings (knowledge adoption)? Do beneficiaries utilize strategies taught in
trainings, such as use of recipes, in their households (knowledge application)? Do beneficiaries
retain this nutritional information over time (knowledge retention)? What are the facilitating
factors and potential barriers for beneficiaries for adoption, application, and retention of
nutrition trainings? Are there spillover effects in the diffusion of nutrition knowledge? What are
participant preferences for structure and content of nutrition trainings?
11.1 Background
The literature on behavior change presents a conceptual framework of stages of change
(Glanz et al. 1994), and has been used to analyze individual conduct in relation to cessation of
addictions such as smoking and alcoholism, as well as in the promotion of healthy lifestyles.
These behavioral stages, also examined in Prochaska et al. (1994), in consideration with the
literature from the fields of psychology on planned behaviors (Ajzen 1991) and self-efficacy
(Bandura 1977, 1997), provide a foundation from which to consider the effect of nutrition
messaging on participant behaviors. The trans-theoretical stages of change (Glanz et al. 1994)
consist of pre-contemplation (unaware, not interested in change), contemplation (thinking about
changing), preparation (making definite plans to change), action (actively modifying habits or
environment), and maintenance (sustaining new habits, preventing relapse). This “trans-
theoretical model” has been used to understand the process of the adoption of healthy diets or
changes in dietary behavior (Kristal et al. 1999). Due to restrictions of time, budget, and privacy,
these assessments are often based on self-perception, rather than direct behavioral observations,
as in the case of this study.
These psychological theories focus primarily on individual-level behavior change. The
broader sociological literature on social networks and knowledge diffusion provides a
complement to individual-level analysis by incorporating the role that family, friends and
community may play in household and individual decisionmaking. Smith and Christakis (2008)
reveal that social linkages play an essential role in determining health outcomes, suggesting
that spillover or multiplier effects occur through networks. The theories have been tested in
studies of obesity, smoking, and emotional states such as happiness, in which social connections
up to the fourth degree of separation may influence individual behavior (Christakis and Fowler
2007). These studies provide insights on how changes in individual behavior may influence
household or community behaviors, and vice-versa. In this manner, interventions may cause
contagion or stimulate information flow through systems, and encourage peer effects.
Definitive evidence is lacking on the best way to encourage behavior change, and on the
most effective program design and implementation. Imdad, Yakoob, and Bhutta (2011), in a
meta-analysis of studies that examined the comparative impacts of food with and without
84
nutrition counseling, found that complementary feeding and feeding with nutrition education
both resulted in improved weight and growth in children, and food with counseling was more
effective than counseling alone. The literature suggests that nutrition information interventions,
also known as behavior change communication, may be more effective if accompanied by
financial or in-kind assistance. Curtis et al. (2001) studied hygiene messaging in Burkina Faso,
and found that while not all behaviors changed significantly, some behaviors showed notable
changes. Certain behaviors may prove easier to adopt than others, and trainings of longer
duration and the use of locally applicable materials tend to be more effective. Other
mechanisms for health information campaigns, such as text messaging, have also been shown to
increase knowledge but not necessarily to change behaviors (Mehran et al. 2011).
A more difficult element to assess is how program design and unique delivery
mechanisms for information may effect change. Theoretical models indicate that information
campaigns may be more cost-effective in affecting the consumption of fruits and vegetables in a
large population than other interventions, such as vouchers or a reduction in value-added tax
(Dallongeville et al. 2011). However, in the promotion of positive behaviors such as adopting a
healthy diet, there are a variety of actions or behaviors that the program aims to change. Among
them are influencing food purchase and consumption decisions, food preparation methods, and
increasing nutrition knowledge. This qualitative study aims to explore the role educational
nutrition trainings play in the acquisition, maintenance, and use of knowledge by program
beneficiaries from the perspective of the beneficiaries themselves.
11.2 Methodology
In order to allow for participation by a diverse selection of beneficiaries, and to comply
with time and budget limitations, focus group discussions (FGDs) were determined to be the
most appropriate method. FGDs were held in Lago Agrio and Tulcán and were led by an IFPRI
facilitator and attended by an observer. The FGDs explored beneficiary perspectives and
preferences on nutrition information sessions, and examined potential behavior change in terms
of use of recommended recipes or nutritional practices. Further, participants were queried how
they perceived the transfer to interact or facilitate behavior change and if information was
shared or disseminated through social networks. However, it is important to note several
limitations to this approach. Discrepancy may occur between self-assessment measures of
stages of change and actual dietary practices (Brug, Glanz, and Kok 1997). It is also not possible
to quantify which delivery mechanisms or messages were most effective, although we were able
to capture participants’ opinions and preferences.
All individuals participating in the FGDs were former beneficiaries of the transfer
program. The strategy for the study was to achieve a sample that reflected the characteristics of
the beneficiary population. FGDs were stratified on transfer modality (cash, voucher, food), and
individual participants were randomly selected from those groups. Two FGDs per treatment
arm were conducted in each of the two urban centers (Tulcán and Lago Agrio), for a total of 12
FGDs. FGDs included, on average, nine individuals (varying from 6–11 individuals) for a total
of 106 participants and discussions lasted from 60 to 90 minutes. Overall, most participants in
the trainings were female, as were the individuals in the FGDs. These women were of different
ages, marital status (many widows and single mothers), and nationality. The majority did not
85
have spouses who attended trainings. While conducting a higher number of FGDs would be
preferable to increase representativeness, a very large sample was not necessary, as saturation
of reoccurring themes will eventually take place. The time line for fieldwork was designed as to
occur approximately 6 months after the last transfer disbursement, or in March 2012.
Audio recordings were collected on two handheld devices with the full knowledge and
permission of participants. The facilitator and observer team completed two focus groups a day,
with participant outreach and scheduling arranged beforehand by WFP-CO. Discussions were
then transcribed for NVivo coding and analysis.
11.3 Results
Beneficiary Preferences: Trainings and Transfer Type
Beneficiary response to the trainings was positive, with participants appreciative of the
location, format, and content of the trainings. The location of the trainings was broadly
perceived to be convenient, often located within or close to the community. An exception
occurred in one of the FGDs in Lago Agrio, where several people complained that the trainings
were far from their homes.
“They explained things to us…we joked and played games. I learned many things, and as I have a
grandson, I will teach my daughter-in-law” — Older Ecuadorian woman, Lago Agrio, cash recipient.
“We have to do our domestic chores, take care of things, the children, so they [the trainings] came
at the exact time when one has a moment, has a free hour” — Colombian woman, Tulcán, cash recipient.
In relation the timing and frequency of trainings, beneficiaries were positive and
appreciated that in some cases they were able to choose between a morning and an afternoon
session, and thus able to complete their other responsibilities. Many participants were stay-at-
home mothers, and found trainings to not be overly burdensome. Furthermore, for some
women, trainings represented a break from routine where they could relax from their daily
responsibilities. On the other hand, a handful of working female participants suggested
weekend sessions would be preferred because of scheduling conflicts. The WFP trainers were well-received, both in terms of the content presented and
attitudes toward participants, who shared the materials in a clear manner, even if that content
was unfamiliar. Furthermore, beneficiaries stated that they generally did not have access to
other sources of nutritional information, and that any available courses on the subject were
costly. Internet connectivity and recipe books were also perceived as expensive or inaccessible
luxuries. Beneficiaries reported that they were exposed to ideas and knowledge in these
sessions. Techniques for food preparation and improving diet as learned in the trainings were
seen as useful to those who attended.
Participants expressed interest in continued exposure to information, both in terms of
further trainings, and not solely for the continuation of the transfers. Furthermore, the addition
of new thematic areas was requested, among these several unrelated to nutrition; psychological
support for Colombians, nondiscrimination, domestic violence and issues on machismo, access
to services and work opportunities, and nutritional practices for the elderly. Participants related
86
a desire for more interactive games such as the bingo, which proved extremely popular. One
participant noted that games or exercises facilitated knowledge transfer for those beneficiaries
who are not literate and thus cannot understand written materials. However, more support was
needed in learning how to prepare the recipes provided by the program, particularly in form of
practice, or demonstrations during trainings, and that would allow for testing of the difficulty
level as well as the resulting flavors of the recommended dishes.
“I come from a farm where nobody has knowledge. In these chats [trainings] one learns to separate
foods, of how one should eat” — Colombian woman, Lago Agrio, cash recipient.
“I am poor but there are people poorer than I, people who sometimes do not eat.” — Ecuadorian
woman, Lago Agrio, cash recipient.
However, some dissatisfaction was expressed with various components of the program.
One general criticism was around the targeting of the program, in that it did not reach the most
vulnerable; and that some that received the benefit were not in sufficiently poor to qualify. In
addition, beneficiaries related disappointment and anxiety over the termination of the transfer
program. Difficulties were recounted in adjustments to household budgets without the transfer.
Some beneficiaries noted that with the end of the program, they reverted back to their former
purchasing patterns and potentially former behavior patterns as well.
“I had a neighbor who had hurt himself, and he couldn’t work, so I sent him a bag of lentils and
rice so he could eat it with his children” – Ecuadorian woman, Lago Agrio, food recipient.
“We received the food, the lentils, we can sell [the food] to buy whatever we are lacking” — Colombian woman, Lago Agrio, food recipient.
A number of modality specific opinions and experiences were found. In general, echoing
findings from the quantitative survey, beneficiaries expressed preference for the modality
received during the program. For example, food recipients believe that with the voucher one
receives less food, in contrast to the food transfer, which is easily stored and saved. For many
food beneficiaries, the items in the food basket lasted after the end of the program, and leftover
foodstuffs, could be shared with relatives or friends. Food could also be re-sold or traded if
necessary. Food recipients noted that with the cash modality some cards were blocked, and that
the food basket does not allow unhealthy or unwise food choices that could be made with cash
purchases.
Cash beneficiaries saw the voucher and food transfers as inflexible, while cash permits
for planning and saving towards other needs. Cash may also be used for nonfood necessities
such as clothing. Supermarkets that redeem vouchers are perceived as expensive, and that cash
holds value better as market prices are lower and cash may be used anywhere without
limitation. Additionally, beneficiaries state that markets provide a greater variety of products,
while supermarkets are seen as having low quality produce, and occasionally for lacking certain
products, in particular, meat.
87
“[In the supermarket] with the list [voucher], my friend says they gave the most expensive things.
Five dollars for me is a lot I am saving” – Colombian woman, Lago Agrio, cash recipient.
“With the cash, you can save twenty for any other thing [that may come up]” – Ecuadorian woman,
Lago Agrio, cash recipient.
Despite expressing preference for their modality type, voucher recipients complained
more than other transfer recipients. Voucher recipients cited bad treatment by store employees,
crowding, high prices, a lack of availability of certain products, low quality produce, and
complaints of items not included in the voucher list. Participants from Tulcán were particularly
vocal about their dissatisfaction, and some were even convinced they were being cheated.
Williams et al. (2012) found that low-income women in Australia perceived fruit and vegetables
to be expensive when those items were less available and of lesser quality. This may also be due
to the fact that many women may have previously purchased the same items in the open
market, where produce prices are generally lower.
“This void persists; you got accustomed to receiving it [the transfer] and having that $40 dollars,
which was already planned for” – Colombian woman, Lago Agrio, cash recipient.
“We obtained new products when we received the voucher, but now after the voucher I think we
will get the same old products, from before.” – Ecuadorian woman, Tulcán, voucher recipient.
As the complaints were particularly vehement in Tulcán, and interviews with the
supermarket owner confirmed that receipts were not given to participants, it may be that
beneficiaries perceived the higher cost because of the lack of transparency regarding their
purchases4. Despite the large number of complaints, voucher beneficiaries felt cash transfers
would not be spent appropriately by recipients.
“In the first two months I was very satisfied; I took home 5 full bags, but in the third month I only
got 3. I think they were taking our 15 cents and we couldn’t complain because in the first place we
aren’t paying for it and we didn’t know if the prices were correct.” — Ecuadorian woman, Tulcán, voucher
recipient.
Despite overall satisfaction with the program, it should be noted that participants may
have believed that their responses in FGDs could influence future inclusion in WFP programs,
despite explanation that their responses would not. During and after FGDs, several beneficiaries
4 In a separate interview, management of a Tulcán supermarket acknowledged variety of operational
issues such as problems with customer volume, treatment of beneficiaries by store employees, and a lack
of stock, especially of fresh produce. To address these issues, trainings were conducted with employees,
days for use of voucher were spread out so as to avoid crowding, and increase purchase of the items in
question; however, this also lead to problems with produce spoilage. It was also noted beneficiaries did
not receive receipts, as they were instead sent to WFP to be redeemed.
88
asked the facilitator and observer about whether WFP would continue its programs, even after
being told these individuals were not WFP staff.
“Not only rice with plantains, but lentils and green salad…my son says ‘I love salad’ since I have
been in the trainings” — Ecuadorian woman with formerly anemic child, Lago Agrio, voucher recipient.
“The colorful plate works because children eat one color and then another color; I think this had the
most results” — Ecuadorian woman, Tulcán, voucher recipient.
Behavior Change, Knowledge Acquisition, and Retention
Information collected in the quantitative survey revealed that nutrition knowledge
increased from baseline in all treatment groups in 6 of 8 key questions, and small increases were
also seen in the comparison group. The FGDs allowed for more open discussions of what
participants recalled from trainings, and reaffirmed the findings from the quantitative survey,
showing consistent basic knowledge retention across categories.
The “Colorful Plate”: Dietary Diversity
Program beneficiaries generally understood why one should eat a colorful plate, and
how these elements were related to health. The rhyming slogan utilized in the trainings; “el plato
colorido es un plato nutritivo” (A colorful plate is a nutritious plate) was easily recalled. The
colorful plate may also appeal to children, in that the color catches their attention and can be
seen as a game. Participants understood the importance of using different types of food, as
represented by colors, to increase the nutritional value of their meals. In particular, beneficiaries
grasped the need to include vegetables, fruits and salads to complement traditional proteins
and starches such as rice, lentils and meat. Additionally, beneficiaries acknowledged that that
the common practice of combining carbohydrates with carbohydrates is not advisable, as it is
unhealthy and causes weight gain, among other problems. Instead, they stated one should
substitute another starch with another food group, such as salad. Only a small minority of
participants did not thoroughly understand the rationale behind this message.
“I like to combine food, to have salad, we used to make rice with pasta but that was bad - flour with
flour - which makes you fat” — Ecuadorian woman, Tulcán, voucher recipient.
“In Tulcán we are addicted to rice, pasta, and potato. The most interesting thing they told us is
that we shouldn’t do that – we have to make a colorful plate” — Ecuadorian woman, Tulcán, cash recipient.
Anemia
Most program participants were comfortable in recalling information on anemia,
reflecting an understanding of the condition and the relationship between certain foods and
improved anemia status. There was widespread recognition of the external symptoms; “Anemia
is when one is pallid, with no color, and does not have an appetite”; or when one is
underweight and even lead to death, in some cases. However, only one participant mentioned
the internal symptoms of the condition; “no appetite, and do not have red blood cells.” Many
beneficiaries specifically identified the connection between consumption of iron-rich foods such
as liver and leafy greens and improvement in anemia status.
89
Beneficiaries did, however, emphasize the connection between iron intake, whether
through consumption of certain foods or as supplement, and the use of other micronutrients.
Many beneficiaries mentioned other vitamins as also having an impact on anemia status, for
example, calcium and iodine. On the other hand, one detailed point - the use of vitamin C
(orange juice) to increase absorption of iron - was not well understood. Traditional methods of
anemia treatment were mentioned as potential solutions, such as pigeon blood, liver and
blackberry licuados (blended drink). Several participants had children diagnosed with anemia
after the baseline survey. When asked about what had happened since, the women said their
children had recuperated. These changes were attributed to adaptation to their children’s’ diets,
such as incorporating greens, lentils, meat, and liver. These accounts suggest that in some
households, anemia is being managed through dietary changes spurred by contact with the
health system and nutrition trainings.
“In the first [survey] my daughter had anemia and I went to the clinic, they gave iron and vitamins,
and in the second [survey], she did not“ — Ecuadorian woman, Lago Agrio, voucher recipient.
“I received a lot of lentils in the last distribution, so I ate and gave some to my daughter with
vegetables, thanks to God she did not have anemia any more when they returned” — Ecuadorian woman,
Lago Agrio, food recipient.
“My daughter was very anemic. They told me in the trainings to make a lentil or a chard soup with
liver” — Colombian woman, Lago Agrio, cash recipient.
Pregnancy, Complementary Feeding, and Nutrition for Young Children
Trainings covered a variety of topics concerning the diet of pregnant and breast feeding
women, as well as appropriate care for babies and young children under the age of two years.
These subjects contained information on the progression of complementary foods, including
quantity, types of foods, and preparation as well as the importance and duration of breast
feeding. Hygiene, such as hand washing, and other sanitary behaviors particularly in regards to
food preparation and storage, was also covered.
Beneficiaries showed good knowledge retention for all the general concepts covered in
this category. Content best retained included information on care, with recommendations such
as the prohibition of harmful behaviors like smoking. Also well remembered were behaviors
such as changes in eating habits during the gestation period to support the growth of the fetus.
Beneficiaries recognized the need to be careful about what pregnant woman consume in order
to abstain from harmful substances, and to ensure dietary diversity with fruits and vegetables
in order to provide sufficient micronutrients to the fetus. Some participants also highlighted the
importance of consuming iron-rich foods, or taking vitamin supplements. However, while
participants widely understood the importance of micronutrients, they did not seem to
remember which micronutrient led to which deficiencies, or which foods contained each
micronutrient, with the notable exception of iron.
90
“When a child is sick you should give more liquids, and should not give soda because when they
eat nutritious food they lose the nutrients.” — Ecuadorian woman, Tulcán, cash recipient.
“You can stew lentils and feed it to children, it is a great food” — Ecuadorian woman with anemic
child, Lago Agrio, food recipient.
In terms of the knowledge retention of nutrition information for young children,
participants reliably recalled the recommendation for exclusive breastfeeding until 6 months,
followed by complementary foods, and slowly progressing to regular foods. Other information
recalled included how to treat sick children, and how to ensure a nutritious diet, and also ways
in which certain foods could better be prepared for children’s tastes. The adoption of hygienic
practices like hand washing, regular medical visits, and vitamins were mentioned as important
to child development, as well as ensuring that children are served first.
Overall, program beneficiaries exhibit a range of stages of behavior change. Few
participants appeared to be in a stage of pre-contemplation, or unaware or uninterested in
change. Some participants stated they were still considering the use of new practices (pre-
contemplation), while others had tested recipes or were utilizing recipes on a rotating basis a
handful of times a week (action). Behavior change was reported by beneficiaries across a
number of different activities including food purchase, preparation and consumption, and in
changing food preparation habits to reflect healthier choices.
“Since we began the trainings I totally changed my way of living -with less sugar and salt” — Ecuadorian woman, Tulcán, cash recipient.
“Pureed vegetables, quinoa, rice, we vary [our diet], we don’t make as much fried food.” — Ecuadorian woman, Tulcán, cash recipient.
“It takes the same time to make a salad as it takes to fry potatoes” — Ecuadorian woman, Tulcán,
voucher recipient.
Some women expressed a desire for their husbands to participate to help change their
attitudes towards nutrition. A small handful of women said their husbands had attended an
occasional training, mostly when their wives were not present as an obligation to receive the
transfer. Generally, machismo was mentioned as a problem, as was domestic violence. As one
woman put it, the lack of male attendance was because “all of them are machista.”
“I would like my husband to go to training, but he doesn’t want to go. I told him let’s go so that
you can learn how to combine foods and he said ‘What crap – this lettuce and chard!’” — Colombian
woman, Lago Agrio, voucher recipient.
“He is not interested; he sees the food and looks like an angry bull about to explode” — Ecuadorian
woman, cash recipient.
91
Furthermore, in terms of behavior change, most men were not interested in cooking.
Spouses may be resistant to changes in household cooking practices, potentially generating
conflict within the household. Women may adapt their cooking practices and purchases to
reflect the preferences of their husbands, which often reflect traditional tastes. This resistance to
change is potentially limiting to positive behavior changes in the household diet since spouses
may exert power (whether physical, emotional or otherwise), over their wives. On the other
hand, the women also assert that they still yield decision-making power over determining the
content of household meals. Women did identify exceptions of men who shared in the
household cooking responsibilities, especially in regards to food purchase and preparation.
“[My husband] has to eat because it is already made. They don’t have anything else to eat, because
they can’t cook themselves.” — Ecuadorian woman, Lago Agrio, cash recipient.
“My husband does participate in the kitchen and he knows how to cook.” — Ecuadorian woman,
Tulcán, voucher recipient.
“For example, with chard…I added a bit of sausage and potato, and then they [children] eat it.” — Ecuadorian woman, Tulcán, cash recipient.
There was evidence of changes of purchase patterns across modalities, even among
those who received a food, who said once staples were accounted for they had money on hand
to buy non-staple foods. Purchases reflect the content of trainings; more vegetables and micro-
nutrient dense foods, less fatty foods, and sweets. Still, beneficiaries express preferences for oil,
salt and sugar, as they facilitate all cooking and give flavor. Furthermore, if low on funds, basic
foods are the first priority for purchases, including rice, potatoes, oil and sugar. Beneficiaries
began to accept that food purchasing changes can be economical, and that there are cheap,
nutritious alternatives to existing food choice.
“I can’t make them [recipes] because they require meat, which is expensive” — Ecuadorian woman,
Lago Agrio, voucher recipient.
“There are people who don’t eat many beans. Lentils are cheaper than buying chicken and it’s
nutritious. It’s better if you have meat, but you can’t, I buy lentils” — Ecuadorian woman, Lago Agrio,
cash recipient.
However, customary diets and traditional foods are a difficult obstacle. It is therefore
not evident whether beneficiaries will maintain these changes over time without the transfer,
which eases financial limitations for purchases and consumption.
92
“I have saved [the recipes] but I haven’t put them into practice because it is difficult to get
everything they ask” — Ecuadorian woman, Lago Agrio, food recipient.
“When I don’t have meat, I buy heart, and I put that in the recipes” — Ecuadorian woman, Tulcán,
food recipient.
“The recipe called for white wine but there wasn’t any in the store – so I put in lime juice”
— Ecuadorian woman, Tulcán, cash recipient.
The utilization of specific recipes depended on the availability of the ingredients in the
household, and access to food products. Popular recipes included quinoa or spinach soup,
spinach omelet, noodles with liver, hominy with eggs (motepillo), fish, chicken soup (sopita de
menundencias), sweet beans (dulce de chocho) and black beans and rice. Beneficiaries reported that
recipes were equally or less time-consuming as preparing traditional dishes. Take home
materials were helpful, as the participants mentioned the importance of having the recipes as a
household reference. Some
mentioned the idea of practicing
recipes in trainings as a way to
facilitate acceptance and adoption.
Questions remain of the use of these
materials for those that are illiterate, especially the elderly. It was evident that the oldest
members of the FGDs were the least engaged, and retained less information.
Accounts of recipe use show that beneficiaries are testing and adapting new behaviors.
Most beneficiaries found recipes to be straightforward, although it was not possible to acquire
all ingredients consistently. Ingredient substitution occurred, primarily due to financial
considerations; and resulted in less expensive meals. In addition, adaptation of recipes also
occurred to facilitate acceptance by children, such as blending vegetables or adding other foods
to make “new” foods palatable. One area where behavior change was less evident is for other
health practices like breastfeeding. Attitudes of household members may affect uptake of
nutritional practices, especially by spouses and children. There are differing levels of support
from spouses, some which welcome new behavior, and others which wanted to maintain
previous habits.
Social Networks
Social cohesion affects the flow of information in a community, which in turn affects
social control, or the ability to encourage positive communal behaviors (Entwisle et al. 2007).
Population turnover, due to migration or other causes, affects the quality of social ties, in that
residential stability has a positive effect (Sampson 1991). Cohesion and control form what
Sampson and colleagues (1997) call “social efficacy,” or the willingness of members to act to
improve a community, which can affect violence or other social ills. Social efficacy is influenced
by community networks, which vary in the complexity of social ties, connectivity, and
permeability of boundaries (Entwisle et al. 2007).
“How will this rice turn out, will the lentils fully cook?
If we prepared a recipe in each session…well, I think we
should do it” — Ecuadorian woman, Tulcán, voucher recipient.
93
The study clusters are urban or peri-urban, which may affect the “density of
acquaintanceship” at the community level
(Freudenburg 1986). The factors that affect the
density of acquaintanceship include length of
residence, population size, diversity and
segregation. The communities in the program
also include those with a high proportion of
Colombian refugees, most of who maintain
cross-border relationships with their former
communities. Participants, both Ecuadorian
and Colombian, recounted that they met new people, and that new relationships were formed
in the trainings. In some cases, a sense of solidarity or common objective was fostered within
neighborhoods. Despite this fostering of new relationships, the duration of the trainings may
have been too brief to solidify relationships; or to ensure contact after program closure.
“We lack a bit to get to all get to know each other, there wasn’t confidence between us” — Ecuadorian woman, Tulcán, cash recipient.
“In Colombia when my mother cooks she says ‘neighbor, I want to invite you over because we are
cooking,’ but here that doesn’t happen” — Colombian woman, Tulcán, food recipient.
In addition to ethnicity, there may also be social or cultural norms that affect sharing
practices particularly around nutrition knowledge. Personality may also impact how
individuals receive knowledge or share information. Rogers (2002) identifies “early adopters”,
who diffuse innovation and try new behaviors. Others may be late adopters, who wait until
other people have already adopted behaviors before trying to change (Bandura 1986). While we
understand this phenomenon occurs, less is known about the “geography” of dissemination.
Fowler and Christakis (2008) describe a “contagion” of emotional states dependent on
geographic proximity, affecting clusters of individuals through relationships and social ties.
There was evidence of information sharing, both from by word of mouth and from
materials. Conducting trainings by barrios may be useful because ideas may “spread” due to
physical proximity and extant social ties, and may increase the likelihood that those who did
not participate will be “exposed” through the diffusion of knowledge. These results are
consistent with the results from the survey, in that 89 percent report of beneficiaries reported
sharing information learned with friends or neighbors. Nutrition knowledge also increased for
the comparison group households, suggesting that beneficiaries may share knowledge and
there is a beneficial spillover effect.
“I have been able to share [what I learned] with friends and young people: you should nurse your
baby until 2 years. These trainings were very important” — Ecuadorian woman, Lago Agrio, cash recipient.
“With my neighbors…they asked to borrow the recipe, they wanted to learn; they asked me to make
copies.” — Colombian woman, Lago Agrio, voucher recipient.
We didn’t know each other, but after, little by
little, we became like a family” — Ecuadorian
woman, Tulcán, cash recipient.
“We made friends with the other women; they
were from all parts of Colombia, from Guayaquil,
from everywhere” – Ecuadorian woman, Lago Agrio,
food recipient.
94
Discrimination
According to beneficiaries, Colombians face difficulties in Ecuador, from discrimination
in social and work environments due to refugee status. On the other hand, one FGD in
particular unearthed resentment for Colombians as they are perceived as having more outlets
for assistance, including UNHCR, IOM, and the Red Cross and local foundations. Ecuadorians
feel they are excluded; Suspicion and dissatisfaction exists regarding the targeting for assistance
programs, both in terms of nationality and by perceptions of poverty. Thus the food, cash and
voucher program was praised in its inclusive approach to reach both vulnerable groups. This
finding is consistent with the quantitative results presented in Chapter 7 which show decreases
in experience of discrimination for both Colombians and Ecuadorians.
“It is hard to find work when you are Colombian and uneducated - they reproach you.” — Colombian woman, Lago Agrio, cash recipient.
“At the foundation they gave food but when I went…they said it was only for Colombians.” — Ecuadorian woman, Lago Agrio, cash recipient.
11.4 Qualitative Conclusion
Qualitative evidence shows beneficiary response to the trainings was positive, including
interactions with WFP staff. In addition, beneficiaries are retaining information, sharing with
their social networks, and occasionally incorporating recipes into their diet. The program may
also have positive social effects in including both Ecuadorians and Colombians in the program.
It is unclear from the limited evidence the scale at which this phenomenon may occur, and
whether the duration of the training period was long enough to affect these intra-community
relationships. This program has widely avoided the resentment generated by programs that
only target one nationality.
“We are very grateful for this program that included not only Colombian families but also
Ecuadorian, because while the government attends to refugees, there are also many poor Ecuadorians
too” — Ecuadorian woman, Lago Agrio, cash recipient.
Two remaining areas of concern for the program include the support of household
members, primarily spouses, and the maintenance of habits without the benefit of the transfer.
Prochaska et al. (1994) posits that interventions should increase the pros of behavior change, the
receipt of a transfer that alleviates financial pressures, and decreases the cons. Incorporating
male members of the household (especially spouses) into the training process could help
decrease a potential barrier to change. Another issue is complaints from beneficiaries that they
needed more preparation to accurately prepare new dishes. Recipe testing, or “demonstration
of diet”, was utilized with success in two notable behavior changes studies in Bangladesh (Roy
et al. 2005) and in Malawi (Hotz and Gibson 2005), both of which found significant impacts of
nutrition education on nutrition outcomes. This practice could be especially appropriate for
illiterate beneficiaries. In general, literature suggests that trainings appear to be a useful
complement to the transfer program; however, the utility without the transfer remains
95
unproven. Nutrition education is not generally viewed as effective as a stand-alone intervention
without a complementary transfer or other component. However, a handful of recent studies
provide initial evidence of the impact of nutrition trainings without a complementary
intervention (Roy et al. 2005), as well as in the case of an infant nutrition education program in
Malawi (Fitzsimons et al. 2012), where consumption and diets improved for young children as
well as in spillover effects to older children.
96
12. Impact Evaluation Conclusion
This report evaluates the impact of the WFP’s Food, Cash, and Voucher intervention in
Northern Ecuador on outcomes including food security, social capital, anemia, and gender
issues. In addition, to the impact assessment, this report provides evidence surrounding
participants experience with the program, conducts a costing study to examine which modality
(food, cash, or voucher) provides the greatest benefit for the amount of funds invested, and
presents the results from a qualitative study on the efficacy of the nutrition trainings that
accompanied the transfers. Moreover, because the program targeted Colombian refugees and
poor Ecuadorians in Northern Ecuador, conclusions and policy implications are also examined
by nationality. This evaluation study in Ecuador is one of several impact evaluations being
undertaken in different countries by the WFP and IFPRI in which various forms of transfers are
compared to learn which modalities are most effective in different contexts.
The results from this report lead to conclusions and policy implications that can be
divided into three main components: program impacts (or benefits to participants), participant’s
costs and preferences, and program costs.
Program Benefits
Overall, program participation leads to large and significant increases across a range of
food security measures, with the value of per capita food consumption increasing by 13 percent,
per capita caloric intake increasing by 10 percent, HDDS improving by 5.1 percent, DDI by 14.4
percent, and FCS by 12.6 percent. The program also leads to a significant decrease in the
percentage of households in the sample with “poor to borderline” food consumption scores by 4
percentage points (or a total percentage reduction of approximately 40 percent). Although all
three modalities improve the value of food consumption, caloric intake, and dietary diversity
measures, vouchers lead to the largest gains in dietary diversity and food leads to the largest
increase in caloric intake. Consistent with the results on food security measures, vouchers lead
to significant increases in the largest number of food groups, increasing consumption in the last
7 days among 9 distinct food groups, while these numbers are 5 and 7 for food and cash
modalities, respectively. Both Colombians and Ecuadorians benefit from participating in the
program; however, Colombians in the food and cash groups experience significantly greater
gains in dietary diversity as compared to Ecuadorians.
Overall, program participation increases social capital among beneficiaries.
Discrimination decreases by 6 percentage points and participation in groups increases by 6
percentage points. Although only cash leads to a significant decrease in discrimination, the size
of the impact is not different across treatment arms. Cash also leads to a significant increase in
trust of institutions and decrease in trust of individuals. The increase in trust of institutions is
significantly different to that of vouchers and the decrease in trust of individuals is significantly
different to that of food. On the other hand, only vouchers lead to significant increases in
participation in groups, and the size of the impact is significantly different to that of cash.
Program impacts on social capital vary across nationality, with treatment leading to larger
impacts on participation for Colombians than for Ecuadorians.
Overall, participation in the program does not lead to changes in household
decisionmaking or experience of disagreement regarding decisionmaking across a wide range
97
of social and economic domains. The program does, however, lead to a significant decrease in
intimate partner violence. Across modalities, there are no significant differences in the size of
the impact for any of the decisionmaking or intimate partner violence indicators and there are
no differential impacts with respect to being Colombian.
Finally, overall participation in the program does not lead to a significant change in
hemoglobin levels or anemia classifications for either children aged 6 to 59 months or for
adolescent girls aged 10 to 16 years. The program does, however, lead to a significant decrease
in hemoglobin levels and increase in anemia for children in the food group. In addition,
moderate or severe anemia increases among children in the cash group. When interacted with
nationality, the increase in anemia among children in the food group is concentrated among
Colombian households. In contrast, there are no significant impacts by either treatment arm or
nationality for adolescent girls.
Participant’s costs and preferences
Although vouchers have the largest impact in terms of food security, voucher
participants report having the most difficulties with the program, with 79 percent of voucher
beneficiaries reporting at least one complaint with their transfer, compared to 40 percent in the
cash group and 37 percent in the food group. Common difficulties experienced by the voucher
recipient households include high food prices and lack of food in supermarkets, as well as
problems at cashiers and understanding rules or use of vouchers. In general, cash is the
preferred modality among participants. In terms of opportunity costs and transportation costs,
cash is the least costly, which is consistent with participants’ revealed preferences. Despite
modality-specific complaints, overall beneficiaries report overwhelming satisfaction with the
program, including nutrition trainings and interactions with program staff.
Program costs
The most costly modality to implement from the institutional perspective for WFP was
the food transfer, while cash and vouchers had similar costs. In particular, as measured on a per
transfer basis using modality-specific costs, it cost $3.03 to transfer $40 in cash to a beneficiary,
$3.30 to provide them with a $40 voucher, and $11.50 to transfer $40 worth of food. The
difference in cost between the food ration and the other modalities was primarily due to added
storage, distribution, and contracting. In contrast to the other transfers, food rations require a
degree of manipulation, in that they must be packaged (thus requiring labor), transported
(human resource and transport cost), and stored in warehouses (rental and upkeep costs).
Combining impacts with costs to compare the cost-effectiveness across modalities for food
security measures, we find that food vouchers are the most cost-effective means for improving
food security and food is the least cost-effective means of improving these outcomes.
Discussion
It is important to emphasize that the results from this evaluation represent a nutrition
knowledge and transfer “bundled” intervention and are specific to the population studied: poor
urban households in Northern Ecuador. Although we find large overall program impacts, these
findings may not hold in rural areas where supermarkets are more likely to have less variety of
foods and may also lack the capacity to properly implement such a program. While these
98
results cannot be generalized to rural areas, they do provide insight for a large, vulnerable
segment of the population, the urban poor, and demonstrate how transfer programs have a
short-term positive impact on food security and social relations within households and
communities.
In the context considered here, choosing the “winner” among the different modalities
depends on the objectives of the policymakers. If the objective of these transfers is simply to
improve welfare, cash is preferable. Cash is the modality that beneficiaries are most satisfied
with, and it is the cheapest means of making transfers. Given the budget available to WFP for
this project, shifting from food to cash could have increased beneficiary numbers by 12 percent.
Moreover, cash allows for savings that helps households smooth their food and nonfood
consumption. If the objective is to increase calories or dietary diversity, vouchers are the most
cost-effective means of doing so, followed by cash. Although, vouchers are the most cost-
effective means of increasing caloric availability and diet quality, it is the modality least
preferred by beneficiaries. Thus policymakers are faced with the trade-off of improving overall
welfare or improving specific outcomes. The former gives aid recipients’ autonomy, while the
latter restricts their choices in order to achieve specific objectives.
100
Appendix 2: Costing Methods and Definition
Costing Glossary:
Cost per beneficiary: the total cost of one beneficiary in any modality
Cost per transfer: the total cost to deliver one $40 transfer in any modality
Human resource cost: Those costs originating from staff time, reflected as a hourly
representation of salary
Modality specific cost: Those implementing costs which are unique to one modality and not
applicable to any other modality, for example, the distribution of food or the liquidation of
vouchers.
Non-modality specific cost: Costs of implementation that are allocated across all modalities, in
that they are not specific to one modality type or another (i.e. a cost that cash, food and voucher
all share). However, not all modality specific costs are divided equally; this is to reflect the
differing number of beneficiaries per modality type. Therefore, some non-modality specific
costs are allocated proportionally in relation to the number of beneficiaries per modality type.
Physical resource cost: These are costs that are not related to personnel time, including all
physical materials used in implementation, including vehicle costs, infrastructure or rental
costs, and materials such as paper, ID machines, etc.
Total cost: All implementation costs including non-modality specific and modality specific costs
Introduction
The International Food Policy Research Institute (IFPRI), in conjunction with the World
Food Programme (WFP), conducted a large-scale impact evaluation to examine vouchers, food,
and cash-based transfers. The evaluation provides evidence to help determine the relative
benefits of alternative modalities to the traditional food transfers. IFPRI has developed a costing
protocol to track comparative costs relative to program modality and to assess cost-effectiveness
across the countries involved in the WFP impact evaluation, Ecuador, Niger, Yemen, and
Uganda.
The costing protocol builds off the activity-based costing ingredients method (ABC-I).
Traditional accounting methods do not take into account the opportunity cost of program
activities, or benefits sacrificed when resources are allocated elsewhere. Therefore, accounting
costs often underestimate the true overall cost of program operations. The use of the ABC-I
method allows for opportunity costs, quantified as economic costs, to be included in the total
program costs. This method also allows for the incorporation of “off-budget” expenditures, for
example, donated goods or services that otherwise would not be included as program operating
101
costs. In this case, donated commodities would be incorporated, even if the actual cost of those
rations were provided by donor governments.
This costing report aims to answer the following research questions with the goal to aid
the WFP in decisionmaking on program activities and funding:
1. What are the relative costs of each modality (cash, voucher, and food)?
2. Which modalities are the most cost-effective?
Methodology
In the ABC-I approach, costs are organized into their respective sectors, known as cost
centers. The ABC-I method is a combination of activity-based accounting methods with the
“ingredients” method, which calculates program costs from inputs, input quantities, and input
unit costs (Fiedler, Villalobos, and de Mattos 2008; Tan-Torres Edejer et al. 2003). As the
ingredients method alone does not allocate costs according to program activities, it would not
allow for comparison between modalities. However, this method, when paired with the ABC-I
approach, matches activities with all their corresponding inputs into cost centers.
In typical program accounting, the general ledger of total funds spent may serve as a
reference point from which to detail activities and ingredients (Canby 1995). However, as WFP
program staff currently utilize cost-accounting methods that aggregate data by cost centers that
are not separated by program modality, it is necessary to re-organize and cost these inputs
using the ABC-I approach. Furthermore, as the WFP Executive Board noted in a recent meeting,
“The practice of embedding non-commodity activities in the commodity-based cost structure
results in non-commodity inputs not being properly defined and categorized. This creates
significant difficulties in planning, controlling, managing and implementing such
activities…[and] in benchmarking across projects, developing performance metrics, and
evaluating the impact” (WFP 2010); it may also be necessary to categorize these costs as
recurrent or start-up costs in order to facilitate future data analysis.
A cost profile analysis provides the information with regards to the comparative benefits
of program modalities, as it allocates costs to their subsequent activities and individual inputs
(“ingredients”). Additionally, program costs should be separated into initial program start-up
costs versus recurrent costs. A cost-effectiveness analysis is a further step in evaluating the
various program modalities. As some outcomes may vary according to the program design in
each country, common outcomes should be identified that can be compared across contexts, if
possible. The calculation of cost per unit of desired impact would then be compared across
program modalities for these selected outcomes.
The bottom-up approach of ABC-I is a thorough, albeit time-intensive, method. Detailed
input data are required, such as the percentages of time allocated for personnel costs by activity.
This particular “ingredient” requires the direct participation of country program staff that may
already have many demands on their time. In-country interviews with program staff are
necessary to accurately delineate these types of data. Costs incurred by collaborating partners
for implementation, as well as for other activities, such as the initial census for beneficiary
identification (conducted by survey firm CEPAR), are included so as to properly reflect the
actual cost of delivery of each modality.
102
Process
At several points throughout the evaluation period, IFPRI staff met with WFP program
personnel in Quito to explain the costing methodology and obtain internal procurement and
operations documents. Interviews and information in Quito were primarily conducted with the
CO Head of Finance Unit, CO Head of Operations, and other program staff involved in the
design of the nutrition components of the intervention.
In addition to providing information via interviews and providing existing
documentation, the WFP-CO undertook a brief survey of all staff involved in the program to
determine the percentage of human resource hours dedicated to this project. The spreadsheet
proved extremely helpful in determining the human resource costs. However, as costs were
organized by staff member rather than by activity, a separate spreadsheet was developed in
accordance with the ABC-I methodology and circulated to other WFP-CO staff for comments on
its accuracy.
One issue discussed with CO staff was how to reflect differences between budgeted
costs, or those costs as reflected in the official WFP accounting ledger, and actual costs incurred
in the field or by collaborating partners. In order to accurately include all costs, it would be
necessary to input contracting costs to WFP with partners, not simply the payment to partners
to execute the program as detailed in PRRO agreements. However, if one were to insert both
types of costs for partners, double-counting would occur. Therefore, a separate analysis was
conducted to determine the cost WFP paid for project execution, to contrast with a calculation
for what was actually spent by partners. This parallel analysis may be of interest to WFP as it
gives an idea of whether it is more fiscally sensible to continue to work through local partners
instead of through direct implementation by WFP sub-offices. To ensure that costs of
implementation were accurately reflected, costing interviews were conducted with
collaborating partners for food distribution. Furthermore, interviews with supermarket owners
participating in the voucher scheme were conducted in both Sucumbíos and Carchi. Finally, an
interview was also conducted with the manager in charge at the bank in Sucumbíos. Additional
interviews were conducted by telephone with local partners in Carchi as needed to confirm cost
information provided by central offices.
Costs Explained
Before presenting the results of the costing exercise, it is helpful to provide a few
explanatory notes regarding categorizations and calculations of costs. In relation to the
agreements with collaborating partners, to reflect actual costs incurred, only the cost of
contracting was included rather than the total amount paid by WFP to partners. Furthermore,
these contracts were developed as part of the PRRO, an umbrella program that provided the
food component of the transfer program, yet also includes the provision of rations for other
PRRO activities. In order to assure that the cost excluded those activities that support the PRRO
rather than just the transfer program, a proportional amount was calculated of contracting cost
in relation to the total amount of metric tons of food for the transfer in comparison with total
tons for the PRRO. Variance in the cost of the WFP food basket over time was not included in
this analysis, as similar changes in cost for other modalities due to inflation or other reasons
were also not calculated. However, it is interesting to note, perhaps for future analyses, that the
103
average cost of the donated food basket increased from approximately $40 to $58 per ration
over the intervention period.
The construction costs of storage facilities were not included, primarily because in all
cases with food storage, these facilities existed prior to the commencement of the transfer
program activities. Instead, facility rental values for the intervention period were included.
Other costs related to facilities included renovations or repairs necessary to prepare warehouses
for the receipt of rations. These costs were calculated utilizing the percentage of total tonnage
stored in each facility that were transfer program rations. As an example, in the case of the local
food distribution partner in Lago Agrio, approximately a $1,500 investment in repairs was
necessary for the existing warehouse. However, as those repairs also apply for other programs
implemented by the local partner, a proportional amount of that cost was allocated to the
transfer program costs ($500, or one-third of total weight of metric tonnage stored). The local
partner in Sucumbíos only had one warehouse in Lago Agrio, and thus, the Shushufindi food
delivery arrived direct from Quito. Although there were no storage or facility costs for food
distribution in Shushufindi, the transport cost was $150 (including drivers and gas). In terms of
human resources, staff had already received training in storage and distribution, so no extra
training was necessary to prepare for participation in the cash, food, and voucher transfer
program. Similarly, the food implementation partner in Carchi and the calculation for the
monthly storage facility costs in Tulcán was calculated by the metric tonnage the transfer
program rations occupied of the warehouse, which (39.3 percent of tonnage) was $102.18, and
similarly in San Gabriel (100 percent tonnage) was $200, for a total of $302.18.
In terms of the supermarkets, as partners for the voucher scheme, several additional
costs are noted. However, these costs were not included in the cost analysis because they are
private, not public costs and thus not costs incurred by WFP. In Lago Agrio, the supermarket
invested in human resource capacity by hiring extra staff and upgrading the computer system.
The computer upgrades cost $800, and for the additional staff included 4 people for 15 days of
work at $300/month for 6 months. In Tulcán, the supermarket incurred additional costs
principally in terms of extra employees (two extra persons at the registers at a total cost of
$400/each) to handle the increased flow of customers, as well as the installation of a computer
system (total cost of $800) to handle transactions.
As a final consideration, the sixth month of costs for this analysis was extrapolated from
the previous five months’ data for those activities that continued on into the last month of the
program. Mean values for number of beneficiaries over the six month period were utilized to
conduct the cost-per-beneficiary calculation.
Detailed budgeted and actual costs by modality that are used in the analysis can be
found in Table A3.1 and Table A3.2 in Appendix 3.
104
Appendix 3: Detailed Cost Tables
Table A3.1 Detailed budgeted costs, by modality
*Human Resources
CASH
VOUCHER
FOOD
**Physical resources HR* RF** HR RF HR RF
PROJECT DEVELOPMENT
1.0 Project design
1.0.1 Project documentation
1.0.1.a WFP staff 5,320 5,320 5,320
1.0.2 Guidelines development process
1.0.2.a WFP staff 557 595 192
1.0.3 Document approval
1.0.3.a WFP staff
1.1 Identification of participants
1.1.1 Contracting consulting firm
1.1.1.a WFP staff 12 12 12
1.1.1.b Contracting cost 73,520 73,520 73,520
1.1.2 Census design
1.1.2.a WFP staff 3,660 3,660 3,660
1.1.3 Data collection in the field
1.1.3.a. CEPAR
1.1.3.a.i CEPAR staff
1.1.3.a.ii CEPAR staff training
1.1.3.a.iii CEPAR, transport
1.1.3.a.iv CEPAR, materials
1.1.3.b WFP
1.1.3.b.i WFP staff 80 80 80
1.1.3.b.ii Travel 715 820 426
1.1.4 Data processing
1.1.4.a CEPAR staff
1.3 Budgeting for project activities
1.3.a WFP staff 332 367 263
1.4 CD Work Partnerships & Monitoring System
1.4.a WFP staff 2979 3405 2128
1.5 Travel
1.5.1 Other Travel 3508 3389 3736
PROJECT IMPLEMENTATION
2.0 Mobilization
2.0.1 Public relations meetings
105
2.0.2 Launch
2.0.3 Communications
2.0.3.a WFP staff
2.0.3.b Materials (printing)
2.0.3.c Travel
2.1 Establishing partnerships
2.1.1 Presentation of project to partners
2.1.1.a WFP staff
2.1.2 Prepare contracts with supermarkets
2.1.2.a WFP staff 2,228
2.1.3 Meetings & contracts with supermarkets
2.1.3.1 Field visit Sucumbíos
2.1.3.1.a. WFP staff 85
2.1.3.1.b. Travel 884 0
2.1.3.1.c. Legal Consulting 250 250 250
2.1.3.2 Field visit Carchi
2.1.3.2a. WFP staff 85
2.1.3.2b. Transport 884
2.1.4 Prepare contracts with bank
2.1.4.a WFP staff 912
2.1.5 Meetings and contracts with bank
2.1.5.a WFP staff 271
2.1.5.b Contracting with Bank
2.1.6 Contracts with other partners
2.1.6.a Partner A 5,912
2.1.6.b Partner B 8,488
2.2 Logistics coordination for implementation
2.2.1 Revision project design project delivery
2.2.1a. WFP staff time 216 216 400
2.2.1.b Production of ID cards
2.2.1.c WFP Staff 3,046 3,330 2,169
2.2.1.d Travel 2,389 2,731 1,920
2.2.1.d Machine for printing ID 2,066 2,361 1,476
2.2.1.e Companies ID printers 623 712 445
2.2.2 Preparation of virtual bank accounts
2.2.2a. Production of debit cards 13219
2.2.2.b Transfer Bank 215
2.2.2.c Bank staff
2.2.2.d WFP staff 3799
2.2.2.e Travel 774
2.2.3 Supermarket selection
2.2.3.a WFP staff 1374
106
2.2.4 Travel preparation 1118
2.2.4.a WFP staff 74 85 53
2.3 Beneficiary Training
2.3.1 Training and introduction
2.3.1.a. WFP staff 1873 2140 1338
2.3.1.b. Travel 1730 1977 1397
2.3.2 Planning meetings
2.3.2.a WFP staff 2310 2514 1718
2.3.3 Development of Training Materials
2.3.3.1 Design of training materials
2.3.3.1.a WFP staff 1320 1512 1352
2.3.3.1.b National Consultants 6018 6018 6018
2.3.3.1.c Travel N. Consultants 564 645 403
2.3.2.2 Procurement of training materials 2000 2000 2000
2.3.2.2.a WFP staff 1747 1997 1248
2.3.4 Conduct training sessions
2.3.4.a. WFP staff 10529 11650 8287
2.3.4.b. Rental of space 1094 1227 767
2.3.4.c. Equipment 3364 3203 2403
2.3.4.d Travel 339 387 276
2.4 Voucher development
2.4.1 Design of vouchers
2.4.1.a. WFP staff 215
2.4.1.b. Printing materials 582
2.4.1.c. Voucher Provided
2.4.1.c.i. WFP staff 11,060
2.4.2 Voucher Liquidation
2.4.2.a WFP staff 6,857
2.4.2.b Bank Transfer to Supermarkets
2.5 Food Handling
2.5.1 Food storage
2.5.1.a WFP staff 2,341
2.5.1.b Bodega rental monthly
2.5.1.b.i WFP 19,506
2.5.1.b.i Partner B
2.5.1.b.iii Partner A
2.5.1.c Facility repairs and investment
2.5.1.c.i WFP
2.5.1.c.ii Partner B
2.5.1.c.iii Partner A
2.5.2 Rations preparation
2.5.2.a. Ration preparation & packaging 18,764
107
2.5.2.b WFP staff
2.5.2.c Other materials
2.5.2.d Cost of food ration
2.5.3 Food distribution
2.5.3.a. Transport (truck, gas, drivers, etc.)
2.5.3.a.i WFP 4,365
2.5.3.a.ii Partner B
2.5.3.a.iii Partner A
2.5.3.b Partners staff for distribution
2.5.3.b.i Partner B
2.5.3.b.ii Partner A
2.6 Facilities Maintenance
2.6.1 Sub-offices 235 269 168
2.6.1.a WFP staff 1,260 1,440 900
2.6.2 Sub-offices, materials
2.6.2.a WFP staff 1,529 1,747 1,092
2.6.2.b Travel 1,547 1,768 1,249
2.6.2.1 Sub-offices, maintenance
2.6.2.1.a WFP staff 1,008 1,152 720
2.6.2.1.b Implementation 3,252 3,252 3,252
2.6.2.1.c Office Supplies & Materials 621 710 444
2.6.2.1.d General Services 424 484 302
2.6.2.1.e Communications & Services 74 85 53
2.6.2.1.f Repair and Maintenance 45 51 32
2.6.2.1.g Vehicle Maintenance 1,138 1,300 813
2.6.2.1.h Other Expenses and Services 1,236 1,412 882
2.6.2.1.i Safety Materials and
Equipment 2,291 2,619 1,637
2.6.2.1.j Cost of Security Service 222 254 159
2.6.2.1.k Mail & Courier Services 28 32 28
2.7 WFP Staff hiring
2.7.1 Selection Process
2.7.1.1 Announcements publicity 480 480 480
2.7.1.1.a WFP staff 703 803 502
2.7.2.2 Staff attention
2.7.2.2.a WFP staff 314 358 314
2.8 Travel
2.8.1 Travel
2.8.1.a WFP staff
2.8.1.b Other activities, travel
2.9 Execution of payments
2.9.a WFP staff 2794 2604 282
108
PROJECT MONITORING AND
EVALUATION
3.0 Setting up monitoring system
3.0.a WFP staff 2112 2112 2112
3.0.b Materials (programs, etc.)
3.1 Monitoring data basic
3.1.a WFP staff 1014 1078 884
TOTAL COST (TYPE): 56351 117398 76100 109408 47749 151188
TOTAL COST:
173749
185509
198937
TOTAL COST PER TRANSFER:
$22.55
$21.45
$36.15
TOTAL COST PER BENEFICIARY:
$135.32
$128.74
$216.94
Table A3.2 Detailed actual costs, by modality
COSTING: ECUADOR C&V
*Human Resources
CASH
VOUCHER
FOOD
**Physical resources HR* RF** HR RF HR RF
PROJECT DEVELOPMENT
1.0 Project design
1.0.1 Project documentation
1.0.1.a WFP staff 5,320 - 5,320 - 5,320 -
1.0.2 Guidelines development process - - - - - -
1.0.2.a WFP staff 458 - 458 - 458 -
1.0.3 Document approval - - - - - -
1.0.3.a WFP staff - - - - - -
1.1 Identification of participants - - - - - -
1.1.1 Contracting consulting firm - - - - - -
1.1.1.a WFP staff 12 - 12 - 12 -
1.1.1.b Contracting cost - - -
1.1.2 Census design - - - - -
1.1.2.a WFP staff 3,660 - 3,660 - 3,660 -
1.1.3 Data collection in the field - - - - - -
1.1.3.a. CEPAR - - - - - -
1.1.3.a.i CEPAR staff 6,983 - 6,983 - 6,983 -
1.1.3.a.ii CEPAR staff training - - - - - -
1.1.3.a.iii CEPAR, transport 4,560 4,560 4,560
1.1.3.a.iv CEPAR, materials 907 907 907
1.1.3.b WFP
109
1.1.3.b.i WFP staff 80 - 80 - 80 -
1.1.3.b.ii Travel - 715 - 820 - 426
1.1.4 Data processing - - - - - -
1.1.4.a CEPAR staff 1,367 - 1,367 - 1,367 -
1.3 Budgeting for project activities - - - - - -
1.3.a WFP staff 321 - 321 - 321 -
1.4 CD Work Partnerships and Monitoring
System - - - - -
1.4.a WFP staff 2,837 - 2,837 - 2,837 -
1.5 Travel - - - - - -
1.5.1 Other Travel - 3,508 - 3,389 - 3,736
PROJECT IMPLEMENTATION
2.0 Mobilization - - - -
2.0.1 Public relations meetings - - - - - -
2.0.2 Launch - - - - - -
2.0.3 Communications - - - - - -
2.0.3.a WFP staff - - - - - -
2.0.3.b Materials (printing) - - - - - -
2.0.3.c Travel - - - - - -
2.1 Establishing partnerships - - - - - -
2.1.1 Presentation of project to partners - - - - - -
2.1.1.a WFP staff - - - - - -
2.1.2 Prepare contracts with supermarkets - - - - - -
2.1.2.a WFP staff - - 2,228 - - -
2.1.3 Meetings and contracts with
supermarkets - - - - - -
2.1.3.1 Field visit Sucumbíos - - - - - -
2.1.3.1.a. WFP staff - - 85 - - -
2.1.3.1.b. Travel - - 884 - -
2.1.3.1.c. Legal Consulting - 250 - 250 - 250
2.1.3.2 Field visit Carchi - - - - - -
2.1.3.2a. WFP staff - 85 - -
2.1.3.2b. Transport - - - -
2.1.4 Prepare contracts with bank - - - - -
2.1.4.a WFP staff 912 - - - -
2.1.5 Meetings and contracts with bank - - - - -
2.1.5.a WFP staff 271 - - - -
2.1.5.b Contracting with Bank - - - - -
2.1.6 Contracts with other partners
2.1.6.a Partner A
2.1.6.b Partner B
2.2 Logistics coordination for implementation - - - - -
110
2.2.1 Revision project design project delivery - - - - - -
2.2.1a. WFP staff time 216 - 216 - 400 -
2.2.1.b Production of ID cards - - - - - -
2.2.1.c WFP Staff 3,046 - 3,330 - 2,169 -
2.2.1.d Travel - 2,389 - 2,731 - 1,920
2.2.1.d Machine for printing ID - 2,066 - 2,361 - 1,476
2.2.1.e Companies ID printers - 623 - 712 - 445
2.2.2 Preparation of virtual bank accounts - - - - - -
2.2.2a. Production of debit cards - 13,219 - - - -
2.2.2.b Transfer Bank - 215 - - - -
2.2.2.c Bank staff - - - - - -
2.2.2.d WFP staff 3,799 - - - - -
2.2.2.e Travel - 774 - - - -
2.2.3 Supermarket selection - - - - - -
2.2.3.a WFP staff - - 1,374 - - -
2.2.4 Travel preparation - - - 1,118 - -
2.2.4.a WFP staff 74 - 85 - 53 -
2.3 Beneficiary Training - - - - - -
2.3.1 Training and introduction - - - - - -
2.3.1.a. WFP staff 1,873 - 2,140 - 1,338 -
2.3.1.b. Travel - 1,730 - 1,977 - 1,397
2.3.2 Planning meetings - - - - - -
2.3.2.a WFP staff 2,310 - 2,514 1,718 -
2.3.3 Development of Training Materials - - - - - -
2.3.3.1 Design of training materials - - - - - -
2.3.3.1.a WFP staff 1,320 - 1,512 - 1,352 -
2.3.3.1.b National Consultants 6,018 - 6,018 - 6,018 -
2.3.3.1.c Travel N. Consultants 564 - - 645 - 403
2.3.2.2 Procurement of training materials - 2,000 - 2,000 - 2,000
2.3.2.2.a WFP staff 1,747 - 1,997 - 1,248 -
2.3.4 Conduct training sessions - - - - - -
2.3.4.a. WFP staff 10,529 - 11,650 - 8,287 -
2.3.4.b. Rental of space - 1,094 - 1,227 - 767
2.3.4.c. Equipment - 3,364 - 3,203 - 2,403
2.3.4.d Travel - 339 - 387 - 276
2.4 Voucher development - - - - - -
2.4.1 Design of vouchers - - - - - -
2.4.1.a. WFP staff - - 215 - - -
2.4.1.b. Printing materials - - - 582 - -
2.4.1.c. Voucher Provided - - - - - -
2.4.1.c.i. WFP staff - - 11,060 - - -
2.4.2 Voucher Liquidation - - -
111
2.4.2.a WFP staff - - 6,857 - - -
2.4.2.b Bank Transfer to Supermarkets - - - - - -
2.5 Food Handling - - - - - -
2.5.1 Food storage - - - - - -
2.5.1.a WFP staff - - - 2,341 -
2.5.1.b Bodega rental monthly - - - - -
2.5.1.b.i WFP - - - - - 19,506
2.5.1.b.i Partner B - - - - - 1,200
2.5.1.b.iii Partner A - - - - - 1,813
2.5.1.c Facility repairs and investment - - - - - -
2.5.1.c.i WFP - - - - - -
2.5.1.c.ii Partner B - - - - - 500
2.5.1.c.iii Partner A - - - - - 1,596
2.5.2 Rations preparation - - - - - -
2.5.2.a. Ration preparation & packaging - - - - - 18,764
2.5.2.b WFP staff - - - - - -
2.5.2.c Other materials - - - - - -
2.5.2.d Cost of food ration - - - - - -
2.5.3 Food distribution - - - - - -
2.5.3.a. Transport (truck, gas, drivers, etc.) - - - - - -
2.5.3.a.i WFP - - - - 4,365 -
2.5.3.a.ii Partner B - - - - - 900
2.5.3.a.iii Partner A 600
2.5.3.b Partners staff for distribution - - - - - -
2.5.3.b.i Partner B - - - - 4,800 -
2.5.3.b.ii Partner A - - - - 5,444 -
2.6 Facilities Maintenance - - - - - -
2.6.1 Sub-offices - 235 - 269 - 168
2.6.1.a WFP staff 1,260 - 1,440 - 900 -
2.6.2 Sub-offices, materials - - - - - -
2.6.2.a WFP staff 1,529 - 1,747 - 1,092 -
2.6.2.b Travel - 1,547 - 1,768 - 1,249
2.6.2.1 Sub-offices, maintenance - -
2.6.2.1.a WFP staff 1,008 - 1,152 - 720 -
2.6.2.1.b Implementation - 3,252 - 3,252 - 3,252
2.6.2.1.c Office Supplies & Materials - 621 - 710 - 444
2.6.2.1.d General Services - 424 - 484 - 302
2.6.2.1.e Communications & Services - 74 - 85 - 53
2.6.2.1.f Repair and Maintenance - 45 - 51 - 32
2.6.2.1.g Vehicle Maintenance - 1,138 - 1,300 - 813
2.6.2.1.h Other Expenses and Services - 1,236 - 1,412 - 882
2.6.2.1.i Safety Materials & Equipment - 2,291 - 2,619 - 1,637
112
2.6.2.1.j Cost of Security Service - 222 - 254 - 159
2.6.2.1.k Mail & Courier Services - 28 - 32 - 28
2.7 WFP Staff hiring - - - - - -
2.7.1 Selection Process - - - - - -
2.7.1.1 Announcements publicity - 480 - 480 - 480
2.7.1.1.a WFP staff 703 - 803 - 502 -
2.7.2.2 Staff attention - - - - - -
2.7.2.2.a WFP staff 314 - 358 - 314 -
2.8 Travel - - - - - -
2.8.1 Travel - - - - - -
2.8.1.a WFP staff - - - - - -
2.8.1.b Other activities, travel - - - - - -
2.9 Execution of payments - - - - - -
2.9.a WFP staff 2,794 - 2,604 - 282 -
PROJECT MONITORING AND
EVALUATION
3.0 Setting up monitoring system - - - -
3.0.a WFP staff 2,112 2,112 2,112 -
3.0.b Materials (programs, etc.) - - - -
3.1 Monitoring data basic - - - -
3.1.a WFP staff 1,014 1,078 884 -
- - - - - -
TOTAL COST (TYPE): 64,449 49,345 83,700 40,471 67,375 75,344
TOTAL COST:
113,794
124,171
142,719
TOTAL COST PER TRANSFER:
$14.77
$14.36
$25.93
TOTAL COST PER BENEFICIARY:
$88.62
$86.17
$155.64
113
Table A3.3 Modality specific costs (based on actual costs)
*Human Resources CASH
VOUCHER FOOD
**Physical resources HR* RF** HR RF HR RF
PROJECT IMPLEMENTATION
2.0 Mobilization
2.0.1 Public relations meetings
2.0.3 Communications
2.0.3.a WFP staff
2.0.3.b Materials (printing)
2.0.3.c Travel
2.1 Establishing partnerships
2.1.1 Presentation of project to partners
2.1.1.a WFP staff
2.1.2 Prepare contracts with supermarkets
2.1.2.a WFP staff 2228
2.1.3 Meetings and contracts with supermarkets
2.1.3.1 Field visit Sucumbíos
2.1.3.1.a. WFP staff 85
2.1.3.1.b. Travel 884
2.1.3.1.c. Legal Consulting 250 250 250
2.1.3.1 Field visit Carchi
2.1.3.1a. WFP staff 85
2.1.3.1b. Transport
2.1.4 Prepare contracts with bank
2.1.4.a WFP staff 912
2.1.5 Meetings and contracts with bank
2.1.5.a WFP staff 271
2.1.5.b Contracting with Bank
2.1.6 Contracts with other partners
2.1.6.a Partner A
2.1.6.b Partner B
2.2.2 Preparation of virtual bank accounts
2.2.2a. Production of debit cards 13219
2.2.2.b Transfer Bank 215
2.2.2.c Bank staff
2.2.2.d WFP staff 3799
2.2.2.e Travel 773
2.2.3 Supermarket selection
2.2.3.a WFP staff 1374
2.2.4 Travel preparation 1118
114
2.2.4.a WFP staff 74 85 53
2.4 Voucher development
2.4.1 Design of vouchers
2.4.1.a. WFP staff 215
2.4.1.b. Printing materials 582
2.4.1.c. Voucher Provided
2.4.1.c.i. WFP staff 11060
2.4.2 Voucher Liquidation
2.4.2.a WFP staff 6857
2.4.2.b Bank Transfer to Supermarkets
2.5 Food Handling
2.5.1 Food storage
2.5.1.a WFP staff 2341
2.5.1.b Bodega rental monthly
2.5.1.b.i WFP 19506
2.5.1.b.i Partner B 1200
2.5.1.b.iii Partner A 1813
2.5.1.c Bodega repairs and investment
2.5.1.c.i WFP
2.5.1.c.ii Partner B 500
2.5.1.c.iii Partner A 1596
2.5.2 Rations preparation
2.5.2.a. Ration preparation & packaging 18764
2.5.2.b WFP staff
2.5.2.c Other materials
2.5.2.d Cost of food ration
2.5.3 Food distribution
2.5.3.a. Transport (truck, gas, drivers, etc.)
2.5.3.a.i WFP 4365
2.5.3.a.ii Partner B 900
2.5.3.a.iii Partner A 600
2.5.3.b Partners staff for distribution
2.5.3.b.i Partner B 4800
2.5.3.b.ii Partner A 5444
2.9 Execution of payments
2.9.a WFP staff 2794 2604 282
PROJECT MONITORING AND
EVALUATION
3.1 Monitoring data basic
3.1.a WFP staff 1014 1078 884
115
TOTAL COST (TYPE): 8863 14457 25672 2834 18169 45129
TOTAL COST:
23320
28506
63298
TOTAL COST PER TRANSFER:
$3.03
$3.30
$11.50
TOTAL COST PER BENEFICIARY:
$18.16
$19.78
$69.03
Table A3.4 Non-modality specific costs (based on actual costs)
NON-MOD SPECIFIC COST CASH
VOUCHER
FOOD
HR PR HR PR HR PR
PROJECT DEVELOPMENT
1.0 Project design
1.0.1 Project documentation
1.0.1.a WFP staff 5320 5320 5320
1.0.2 Guidelines development process
1.0.2.a WFP staff 458 458 458
1.0.3 Document approval
1.0.3.a WFP staff
1.1 Identification of participants
1.1.1 Contracting consulting firm
1.1.1.a WFP staff 12 12 12
1.1.1.b Contracting cost
1.1.2 Census design
1.1.2.a WFP staff 3660 3660 3660
1.1.3 Data collection in the field
1.1.3.a. CEPAR
1.1.3.a.i CEPAR staff 6983 6983 6983
1.1.3.a.ii CEPAR staff training
1.1.3.a.iii CEPAR, transport 4560 4560 4560
1.1.3.a.iv CEPAR, materials 907 907 907
1.1.3.b WFP
1.1.3.b.i WFP staff 80 80 80
1.1.3.b.ii Travel 715 820 426
1.1.4 Data processing
1.1.4.a CEPAR staff 1367 1367 1367
1.3 Budgeting for project activities
1.3.a WFP staff 321 321 321
1.4 CD Work Partnerships & Monitoring System
1.4.a WFP staff 2837 2837 2837
1.5 Travel
1.5.1 Other Travel 3507 3389 3735
116
2.0.2 Launch
2.2 Logistics coordination for implementation
2.2.1 Revision project design project delivery
2.2.1a. WFP staff time 216 216 400
2.2.1.b Production of ID cards
2.2.1.c WFP Staff 3046 3330 2169
2.2.1.d Travel 2389 2731 1920
2.2.1.d Machine for printing ID 2066 2361 1476
2.2.1.e Companies ID printers 623 712 445
2.3 Beneficiary Training
2.3.1 Training and introduction
2.3.1.a. WFP staff 1873 2140 1338
2.3.1.b. Travel 1730 1977 1397
2.3.2 Planning meetings
2.3.2.a WFP staff 2310 2514 1718
2.3.3 Development of Training Materials
2.3.3.1 Design of training materials
2.3.3.1.a WFP staff 1320 1512 1352
2.3.3.1.b National Consultants 6018 6018 6018
2.3.3.1.c Travel N. Consultants 564 645 403
2.3.2.2 Procurement of training materials 2000 2000 2000
2.3.2.2.a WFP staff 1747 1997 1248
2.3.4 Conduct training sessions
2.3.4.a. WFP staff 10529 11650 8287
2.3.4.b. Rental of space 1094 1227 767
2.3.4.c. Equipment 3364 3203 2403
2.3.4.d Travel 338 387 276
2.6 Facilities Maintenance
2.6.1 Sub-offices 235 269 168
2.6.1.a WFP staff 1260 1440 900
2.6.2 Sub-offices, materials
2.6.2.a WFP staff 1529 1747 1092
2.6.2.b Travel 1547 1768 1249
2.6.2.1 Sub-offices, maintenance
2.6.2.1.a WFP staff 1008 1152 720
2.6.2.1.b Implementation 3252 3252 3252
2.6.2.1.c Office Supplies and
Materials 621 710 444
2.6.2.1.d General Services 424 484 302
2.6.2.1.e Communications & Services 74 85 53
2.6.2.1.f Repair and Maintenance 45 51 32
117
2.6.2.1.g Vehicle Maintenance 1138 1300 813
2.6.2.1.h Other Expenses and Services 1236 1412 882
2.6.2.1.i Safety Materials and
Equipment 2291 2619 1637
2.6.2.1.j Cost of Security Service 222 254 159
2.6.2.1.k Mail & Courier Services 28 32 28
2.7 WFP Staff hiring
2.7.1 Selection Process
2.7.1.1 Announcements publicity 480 480 480
2.7.1.1.a WFP staff 703 803 502
2.7.2.2 Staff attention
2.7.2.2.a WFP staff 314 358 314
2.8 Travel
2.8.1 Travel
2.8.1.a WFP staff
2.8.1.b Other activities, travel
3.0 Setting up monitoring system
3.0.a WFP staff 2112 2112 2112
3.0.b Materials (programs, etc.)
55,586 34,885 58,028 37,636 49,206 30,213
Total cost, non-modality specific
$90,471
$95,664
$ 79,420
Total cost (non-mod) per transfer
$11.74
$11.06
$14.43
Total cost (non-mod) per beneficiary
$70.46
$66.39
$86.61
118
References
Ahmed, A., D. Gilligan, J. Hoddinott, A. Peterman, and S. Roy. 2010. Evaluating Vouchers and
Cash-Based Transfers: Final Inception Report. Washington, DC: International Food Policy
Research Institute.
Ahmed, A., A. Quisumbing, M. Nasreen, J. Hoddinott, and E. Bryan. 2009. Comparing Food and
Cash Transfers to the Ultra-Poor in Bangladesh. Washington, D.C.: International Food
Policy Research Institute. http://www.ifpri.org/publication/comparing-food-and-cash-
transfers-ultra-poor-bangladesh.
Ajzen, I. 1991. “The Theory of Planned Behavior.” Organizational Behavior and Human Decision
Processes 50 (November 4): 179–211.
Alesina, A., and E. Ferrara. 2002. “Who Trusts Others?” Journal of Public Economics 85 (2): 207–
234.
Bandura, A. 1977. “Self-efficacy: Towards a Unifying Theory of Behavioral Change.”
Psychological Review 84 (2): 191–215.
Bandura, A. 1986. The Explanatory and Predictive Scope of Self-Efficacy Theory. Journal of
Social and Clinical Psychology, Special Issue: Self-Efficacy Theory in Contemporary
Psychology, 4: 359-373.
––––––––. 1997. Self-Efficacy: The Exercise of Control. New York: Freeman.
Barrett, C. B., and D. G. Maxwell. 2005. Food Aid after 50 Years: Recasting Its Role. New York and
London: Routledge.
Brug, J., K. Glanz, and G. Kok. 1997. “The Relationship between Self-Efficacy, Attitudes, Intake
Compared to Others, Consumption, and Stages of Change Related to Fruit and
Vegetables.” American Journal of Health Promotion 12 (1): 25–30.
Caldés, N., D. Coady, and J. A. Maluccio. 2006. “The Cost of Poverty Alleviation Transfer
Programs: A Comparative Analysis of Three Programs in Latin America.” World
Development 34 (5): 818–837.
Canby, J. B. 1995. “Applying Activity-Based Costing to Healthcare Settings.” Journal of the
Healthcare Financial Management Association 49 (2): 50–55.
Christakis, C., and J. Fowler. 2007. “The Spread of Obesity in a large Social Network over 32
years.” New England Journal of Medicine 357 (4): 370–379.
––––––––. 2008. “The Collective Dynamics of Smoking in a Large Social Network.” New England
Journal of Medicine 358 (21): 2249–2258.
Cunha, J. M. 2010. Testing Paternalism: Cash vs . In-kind Transfers in Rural Mexico. Discussion
Paper 09–021. Stanford, CA, USA: Stanford Institute for Economic Policy.
Curtis, V., B. Kanki, S. Cousens, I. Diallo, A. Kpozehouen, M. Sangaré, and M. Nikiema. 2001.
“Evidence of Behaviour Change Following a Hygiene Promotion Programme in Burkina
Faso.” Bulletin of the World Health Organization 79 (6): 518–527.
119
Dallongeville, J., L. Dauchet, O. de Mouzon, V. Réquillart, and L. Soler. 2011. “Increasing Fruit
and Vegetable Consumption: A Cost-Effectiveness Analysis of Public Policies.” European
Journal of Public Health 21 (1): 69–73.
Entwisle, B., K. Faust, R. Rindfuss, and T. Kaneda. 2007. “Networks and Contexts: Variation in
the Structure of Social Ties.” American Journal of Sociology 112 (5): 1495–1533.
Fiedler, J. L., C. A.Villalobos, and A. C. de Mattos. 2008. “An Activity-Based Cost Analysis of
the Honduras Community-Based, Integrated Child Care (AIN-C) Programme.” Health
Policy and Planning 23 (6): 408–427.
Fiszbein, A., N. Schady, F. H. G. Ferreira, M. Grosh, N. Kelleher, P. Olinto, and E. Skoufias.
2009. Conditional Cash Transfers: Reducing Present and Future Poverty. Washington, DC:
World Bank.
Fitzsimons, E., B. Malde, A. Mesnard, and M. Vera-Hernandez. 2012. Household Responses to
Information on Child Nutrition: Experimental Evidence from Malawi. CEPR Discussion Paper
No. DP8915. London: Centre for Economic Policy Research.
Freudenburg, W. 1986. “The Density of Acquaintanceship: An Overlooked Variable in
Community Research.” American Journal of Sociology 92 (1): 27–63.
García-Moreno, C., H. A. F. M. Jansen, M. Ellsberg, L. Heise, and C. Watts. 2005. WHO
Multicountry Study on Women’s Health and Domestic Violence against Women: Initial Results
on Prevalence, Health Outcomes, and Women's Responses. Geneva: World Health
Organization. http://www.cabdirect.org/abstracts/20063002089.html.
Gentilini, U. 2007. Cash and Food Transfers: A Primer. Occasional Paper 18. Rome: World Food
Programme.
Glanz, K., J. Brug, and P. Van Assema. 1997. “Are Awareness of Dietary Fat Intake and Actual
Fat Consumption Associated?—A Dutch-American Comparison.” European Journal of
Clinical Nutrition 51 (8): 542–547. http://ukpmc.ac.uk/abstract/MED/11248880.
Glanz, K., R. Patterson, A. Kristal, C. DiClemente, J. Heimendinger, L. Linnan, and D.
McLerran. 1994. “Stages of Change in Adopting Healthy Diets: Fat, Fiber and Correlates
of Nutrition Intake.” Health Education Quarterly 21 (4): 499–519.
Hanlon, J., A. Barrientos, and D. Hulme. 2010. Just Give Money to the Poor: The Development
Revolution from the Global South. Sterling, VA, USA: Kumarian Press.
Hidrobo, M., J. Hoddinott, A. Margolies, V. Moreira, and A. Peterman. 2011. “Impact
Evaluation of Cash, Food Vouchers and Food Transfers among Colombian Refugees and
Poor Ecuadorians in Carchi and Sucumbios: Baseline Report.” International Food Policy
Research Institute, Washington, DC.
Hoddinott, J., and Y. Yohannes. 2002. Dietary Diversity as a Food Security Indicator. Food
Consumption and Nutrition Division Discussion Paper 136. Washington, DC:
International Food Policy and Research Institution.
120
Hotz, C., and R. Gibson. 2005. “Participatory Nutrition Education and Adoption of New
Feeding Practices Are Associated with Improved Adequacy of Complementary Diets
among Rural Malawian Children: A Pilot Study.” European journal of clinical nutrition 59
(2): 226–237.
Imdad, A., M. Y. Yakoob, and Z. Bhutta. 2011. “Impact of Maternal Education about
Complementary Feeding and Provision of Complementary Foods on Child Growth in
Developing Countries.” BMC Public Health 11 (Supplement 3): S25.
Kennedy, G., T. Ballard, and M. C. Dop. 2011. Guidelines for Measuring Household and Individual
Dietary Diversity. Rome: Food and Agricultural Organization of the United Nations.
King, E. M., and A. D. Mason. 2001. Engendering Development through Gender Equality in Rights,
Resources, and Voices. Research Report. Washington, DC: World Bank.
Kristal, A., K. Glanz, S. Curry, and R. Patterson. 1999. “How Can Stages of Change Be Best Used
in Dietary Interventions?” Journal of the American Dietetic Association 99 (6): 679–684.
Labonne, J., and R. Chase. 2011. “Do Community-Driven Development Projects Enhance Social
Capital? Evidence from the Philippines.” Journal of Development Economics 96 (2): 348–
358.
Manchu, M. T., and H. Mendez. 2007. Tabla de Composicion de Alimentos de Centroamerica,
segunda edicion. Guatemala City: Instituto de Nutricion de Centro America y Panama
(INCAP), Organizacion Panamericana de la Salud (OPS).
Malhotra, A., S. Schuler, and C. Boender. 2002. Measuring Women’s Empowerment as a Variable in
International Development. Background paper for the World Bank Workshop on Poverty
and Gender: New Perspectives. Washington, DC: Gender and Development Group,
World Bank.
McKenzie, D. (2012). Beyond baseline and follow-up: the case for more T in experiments. Journal
of Development Economics, 99(2), 210-221.
Mehran, L., P. Nazeri, H. Delshad, P. Mirmiran, Y. Mehrabi, and F. Azizi. 2011. “Do Text
Messaging Interventions Improve Knowledge, Attitude and Practice Regarding Iodine
Deficiency and Iodized Salt Consumption?” Public Health Nutrition, First View article: 16.
PMA (Programa Mundial de Alimentos [World Food Programme]). 2010. Informe de la
Evaluacion a profundidad de la seguridad alimentaria y nutricional en la poblacion Colombiana
refugiada en Ecuador y de la poblacion de acogida. Quito, Ecuador.
Prochaska, J. O., W. F. Velicer, J. S. Rossi, M. G. Goldstein, B. H. Marcus, W. Rakowski, C. Fiore
et al. 1994. “Stages of Change and Decisional Balance for 12 Problem Behaviors.” Health
Psychology (official journal of the Division of Health Psychology, American
Psychological Association) 13 (1): 39–46.
Putnam, R. 2000. Bowling Alone: The Collapse and Revival of American Community. New York:
Simon & Schuster.
Rogers, E. 2002. “Diffusion of Preventive Innovations.” Addictive Behaviors 27 (6): 989-993.
121
Roy, S., G. Fuchs, Z. Mahmud, G. Ara, S. Islam, S. Shafique, S. Akter et al. 2005. “Intensive
Nutrition Education with or without Supplementary Feeding Improves the Nutritional
Status of Moderately-Malnourished Children in Bangladesh.” Journal of Health,
Population, and Nutrition 23 (4): 320–330.
Sabates-Wheeler, R., and S. Devereux. 2010. “Cash Transfers and High Food Prices: Explaining
Outcomes on Ethiopia’s Productive Safety Net Programme.” Food Policy 35 (4): 274–285.
doi:10.1016/j.foodpol.2010.01.001.
http://linkinghub.elsevier.com/retrieve/pii/S0306919210000230.
Sampson, R. 1991. “Linking the Micro- and Macrolevel Dimensions of Community Social
Organization.” Social Forces 70 (1): 43–64.
Sampson, R., S. Raudenbush, and F. Earls. 1997. “Neighborhoods and Violent Crime: A
Multilevel Study of Collective Efficacy.” Science 277 (5328): 918–924.
Smith, K., and N. Christakis. 2008. “Social Networks and Health.” Annual Review of Sociology 34
(1): 405–429.
Subbarao, K., A. Bonnerjee, and J. Braithwaite. 1997. Safety Net Programs and Poverty Reduction:
Lessons from Cross-Country Experience. Washington, DC: World Bank.
Sullivan, K. M., Z. Mei, L. Grummer-Strawn, and I. Parvanta. 2008. “Haemoglobin Adjustments
to Define Anaemia.” Tropical Medicine & International Health 13 (10): 1267–1271.
Tabor, S. R. 2002. Assisting the Poor with Cash: Design and Implementation of Social Transfer
Programs. Washington, DC: World Bank.
Tan-Torres Edejer, T., R. Baltussen, T. Adam, R. Hutubessy, A. Acharya, D. B. Evans, and C. J. L.
Murray. 2003. Making Choices in Health: WHO Guide to Cost-Effectiveness Analysis. Geneva:
World Health Organization.
UNHCR (United Nations High Commission for Refugees). 2010. Global Report 2010: Ecuador.
Geneva. http://www.unhcr.org/4dfdbf5c0.html.
USDA (United States Department of Agriculture). 2010. USDA National Nutrient Database for
Standard Reference, Release 23. Beltsville, MD, USA: Agricultural Research Service,
USDA.
Webb, P., and S. K. Kumar. 1995. “Food and Cash for Work in Ethiopia: Experiences during
Famine and Macroeconomic Reform.” In Employment for Poverty Reduction and Food
Security, edited by J. von Braun, 201–219. Washington, DC: International Food Policy
Research Institute.
Weismann, D., L. Bassett, T. Benson, and J. Hoddinott. 2009. Validation of the World Food
Programme’s Food Consumption Score and Alternative Indicators of Household Food Security.
Washington, DC: International Food Policy Research Institute.
WFP (World Food Programme). 2008. Food Consumption Analysis: Calculation and Use of the Food
Consumption Score in Food Security Analysis. Rome.
122
––––––––. 2010. Resource, Financial, and Budgetary Matters: Financial Framework Review. Executive
Board, Second Regular Session. Rome.
WFP-Ecuador. 2010. Spanish Trust Fund Proposal: Food Assistance (Cash/Vouchers) to Colombian
Refugee Populations. Quito.
––––––––. 2011. Report on the Cash and Voucher Project: Quarterly Report of Monitoring, May-July
2011. Quito.
White, A. G. 2011. In the Shoes of Refugees: Providing Protection and Solutions for Displaced
Colombians in Ecuador. Geneva: United Nations High Commission for Refugees
http://www.unhcr.org/4e4bd6c19.html.
WHO (World Health Organization). 2011. Haemoglobin Concentrations for the Diagnosis of Anaemia
and Assessment of Severity. Geneva.
http://www.who.int/vmnis/indicators/haemoglobin.pdf.
Williams, L. K., L. Thornton, D. Crawford, and K. Ball. 2012. “Perceived Quality and
Availability of Fruit and Vegetables Are Associated with Perceptions of Fruit and
Vegetable Affordability among Socio-Economically Disadvantaged Women.” Public
Health Nutrition 15 (7): 1261–1267.